http://acl.ldc.upenn.edu/P/P03/ ACL 2003 http://acl.ldc.upenn.edu/P/P03/P03-1070.pdf 70 Towards a Model of Face-to-Face Grounding aaai american annual association brennan california cambridge clark cognition cognitive collaboration collaborative communication computational contributing detecting dillenbourg discourse editors fall gibbs giboin grounding horvitz human language linguistics machine matheson meeting menlo miscommunication misunderstanding modelling models multimodal naacl north obligations paek papers park poesio portland press preventing process psychological references referring repairing rules schaefer science symposium systems traum uncertainty university update using utility wilkes working workshop http://acl.ldc.upenn.edu/P/P03/P03-1047.pdf 47 Bridging the Gap Between Underspecification Formalisms: Minimal Recursion Semantics as Dominance Constraints advantage alex alexander algebra algorithm algorithms alshawi althaus ambiguities amsterdam april available based bene between bodirsky bridging broadcoverage carl cation chicago cient colloquium complexity conclusion conference constraint constraints construction copestake crouch dealing deduction denys development distinguished dominance driven duchier eacl encoding english enumerate environment equivalence evaluation flickinger formalism formalisms france from grammar grammars graph have head hole hpsg information interested interpretation introduction ivan joachim johan journal koller lacl lambda language lascarides lnai logic manuel manuscript markus mehlhorn miele minimal model monotonic nets niehren normal obtained pages papers phrase pollard predicate press proc proceedings provably proved ralf readings recursion references related representation representations researchers resources reyle scope sebastian semantic semantics source stanford stefan structure structures sublanguages that thater their thereby thiel toulouse treinen underspeci university unplugged using volume weakly with http://acl.ldc.upenn.edu/P/P03/P03-1037.pdf 37 Parametric Models of Linguistic Count Data aaai account accounting accuracy adaptation advantages against algorithms already always among andrew anthology applications applied approximate ashutosh assumption ated ation attractive based bayes bayesian been bernoulli beta beyond binomial case categorization cation chance church class classi classical clearly closer clustering communication comparing comparison computation computational computationally conclusion conditions conference conventional count daniel data david decline degree dietterich difference discrete displayed distribution distributions document documents dominates domingos early edition either empirical engineering ennis estimates estimation european evaluation event evidence federalist figure forty four frederick frequencies function gale garg given gradient have highest http increased independence independently inference information inter international interpreted jack jian johnson journal kachites kamal kenneth kotz lang language learning least lewis likelihood linguistic linguistics lowe lter machine maximum mccallum meaningful methods michael mixture mixtures model modeling modelling models most mosteller naive natural netnews neural newsweeder nigam noriegas norman note numerically observed optimality outperforms over overall overdispersed oxford pages papers parameter pawitan pazzani peak pedro place poisson prediction preference preferred presented press prior probabilistic provides rapid recommend references replicated results retrieval roth samuel second sensory should simple simulation size speech springer stable standard statistical stephen straightforward studies substantial supervised tasks technology tests text than that theoretical there they thomas toolkit topic trial tried understanding university unless used using variation vocabulary volume wallace webkb well which wiley william with word workshop york yudi zero zeroin http://acl.ldc.upenn.edu/P/P03/P03-1025.pdf 25 Compounding and derivational morphology in a finite-state setting analyse anke anne approach approximating approximation approximations based bernd bettina bius boullier butt calculus cambridge canada centre coling computational constraintbased context contextfree cookbook corner csli deko deling deutsche developer dmor driven edmund einer emmanuel entwicklerhandbuch entwicklung eugenia evans experiments february fernando finite france free fuer gldv grammar grammars grenoble grimley handbook hans head heid institut intermediate international italy iwpt jahrestagung johnson kiefer king komplexer krieger language languages lecture left linguistics madrid maria mark maschinelle melvin miriam montreal nancy nederhof nite notes number parsing pereira phrase pierre practical press proceedings processing publications rank rebecca references regular report research rique roche rter schabes schiller schmid segond spain sprachverarbeitung stanford state statistical structure student studienarbeit stuttgart supertagging system tanja technical technologies thesis tracy transforms trento uberlich ulrich universit using with workshop wortbildungsgrammatik wright writer wurster xerox yves http://acl.ldc.upenn.edu/P/P03/P03-2017.pdf 88 Towards Interactive Text Understanding aaai access according acknowledgements acquiring adapt advice also amount analysis approach argmax artificial aspects authoring automatic automation based berger between beyond boitet briefly brown brun channel choices coling complementing completed compression computational conclusion connections content context controlled conventional corpus corresponding corresponds deep della dialogue different direction discriminative discussions document documents dordrecht draws dymetman eacl editing emnlp entropy epstein estimated estimation exciting extension extracting extraction fanchon feedback finding first following formal formalism foster framework friendly from functional fuzzy gaussier generating generation goes grammar grammatical have here however human improving information inlg input intelligence interactive interfaces intervention introduced inverted isabelle journal kaufmann kluwer knight knowledge lafferty langlais language lapalme linguistic linguistics machine marcu mathematics maximum mediated mention mercer minimal model models more morgan multilingual natural nickerson nilsson noisy noninteractive normalize note options organizing parallel parallels parameter permitting pietra plamondon plane pogodalla point power prediction presently probabilistic probabilities proceedings produce program programming proposed purpose rank ranta realization recently references relation reliable representation representations represents research resist respectively retrieval reversing scott scratch semantic sentence september sigir source speed spoken spring statistical structure student summarization symposium syntax synthesis system target tasks text texts than thanks that then thesis think this those through thus towards training translation translators transtype typetheoretical understanding university user using where while with working workshop world written yamada york http://acl.ldc.upenn.edu/P/P03/P03-2018.pdf 89 A Prototype Text to British Sign Language (BSL) Translation System american associates austin avatar based basis bellugi boston brien british colleagues corina cormier developed dictionary differential doctoral editors emmorey english erlbaum expresses faber functions generation gesture graduate grammaticization hamnosys hanke illustrated indexic interface language lawrence london norwich numerosity pages phonetic processing references referential reilly schmaling school sign signs space system televirtual texax thesis topographic transcription university visicast within workshop http://acl.ldc.upenn.edu/P/P03/P03-1054.pdf 54 Accurate Unlexicalized Parsing abstract acoustical advances advantages already also ambiguity america amnesia andrew another applying argue arti aspects attachment attachments attempt automaton baker bank based basic been bene best bigram bilexical booth bounds bresnan brought cambridge capture captured caraballo certain charniak chart chomsky christopher cial cient clear climb closure collins communication compact competence component computational computers conclusion conference conjunctions contained context corpora corpus could course cumulative dana daniel david decision dependencies depth descendant development dismal distance dominate dominated donald driven eacl early easier easy edge editor editors eisner emnlp empirical enough entropy estimate eugene explicit feature features figure final ford free further generally gildea giorgio goldwater grammar grammars grammatical gures have head help here higher hill hindle however ieee immediate improve information inspired intelligence intervention irrelevant james jason joan john johnson kaplan kaufmann klatt klein language languages large least lexical lexicalization lexicalized likelihood linear linguistic linguistics lower magerman manning marilyn mark marked marking marks mats maximum measures meeting mental merit methods michael model models morgan naacl naftali national natural necessity neural noam node nodes ofthe orthogonal others overstated pages papers parse parser parsers parsing pcfg penn pennsylvania performance perhaps periphery power prepositions press presumably probabilistic probability proc proceedings processing proxy radford rather recognition references relations rendered representation representations require results richard right rightmost ronald rooth satta section selection sharon shown shows singer sixth society some space speech state statistical statistics structural structure surprisingly syntactic syntax systems taylor test than that their theoretical theory these thesis this thomson those though time tishby took trainable transactions transformational tree treebank trends tried unannotated univ university unlexicalized used variation verbal very volume want ways well which with wolf word words workshop worthwhile yoram http://acl.ldc.upenn.edu/P/P03/P03-1005.pdf 5 Hierarchical Directed Acyclic Graph Kernel: Methods for Structured Natural Language Data analysis answering approach arch attribute automatic based baseline better both cambridge cascaded cation chunking cient classi coling collins communication computational conclusion conference conll convolution cosine cristianini cruz dependency discrete duffy editors entity evaluated evaluation experiments feature framework function generalized goitaikei handling haussler hayashi hdag hirao ikehara indexing information inside international introduction isozaki iwanami japanese journal kazawa kernel kernels kudo language learning lexicon linguistics lodhi machine machines maeda matsumoto measure method methods miyazaki model multidocument nakaiwa named natural nature neuron nical offers ogura other oyama pages paper parsing performance present press problems proc proce proposed publishing question recognition references report rese retrieval richer salton santa saunders semantic shawe shirai showed similarity single space springer ssing statistical string structure structures summarization support system taylor tech technical text texts than that theory this tice track trec university using vapnik vector very volume voorhees watkins which with within wong yang yokoo http://acl.ldc.upenn.edu/P/P03/P03-1008.pdf 8 Syntactic Features and Word Similarity for Supervised Metonymy Resolution abduction ablex accurate agreement annotation appelt argument arti assessing based briscoe cambridge carletta carroll cation cial classi coercions colingacl computational computers constraint copestake corley corpora corpus crocker database deriving electronic extension fass fellbaum finding from general gildea grammar gsearch harabagiu hobbs humanities intelligence interpretation journal kappa keller language lexical linguistics lrec martin mass metaphor metonymic metonymy natural necessity pages palmer parsing polysemy predicate press proc processing productive query recognition references robust rules semantics semi sense stanford statistic statistical stickel structure syntactic system systems tasks text trewin unparsed usage wordnet workshop http://acl.ldc.upenn.edu/P/P03/P03-1030.pdf 30 Optimizing Story Link Detection is not Equivalent to Optimizing New Event Detection accurate allan ambiguity analysis anaphora asymmetric athens australia ayman based bounds brants brian brown bruce cambridge carbonell center charles chen chun cikm coincidence computational conference cooccurrence corpora corpus croft cronen daniel database deguzman delos detection detections diego digital document dunning edmonton edward effect ellipsis enabled enhancement evaluation event farahat feedback finland first flannery francine greece hard hema hopkins hubert information international ioannis jaime james jian jinxi johns jrvelin knowledge kraaij laflamme language larvrenko latent lavrenko libraries line linguistics link lrec malin management martin mclean meeting melbourne methods modeling models multilingual naacl novelty numerical ofspeech pages part patterns personalisation personalization perspective pierce pirkola pohlmann pollard press probabilistic proceedings processing proximity query raghavan rajman ralf recall recipes recommender reduce references relevance renee report research resolution resources retrospective saul searching segmentation semantic sigir slides speech statistics stemming stephen story study successful summer surprise swan systems tampere teukolsky text thomas thorsten timelines topic townsend tracking transactions tsochantaridis umass university using variants veera vetterling victor vienna viewing wayne wessel william with word workshop yang yiming zhang http://acl.ldc.upenn.edu/P/P03/P03-2029.pdf 100 Word Sense Disambiguation Using Pairwise Alignment accuracy acids acquired adam also ambiguity analysis analyzer andrew annual apple assumes baifukan based because beckwith become being between biological biology calculated cambridge center charniak christiane christopher clue computational conference consequently context corpora correct database dekang dependencies dependency derek derived disadvantage disambiguation durbin eacl eddy eugene evaluating evaluation exercise fact fellbaum former foundations from full general genom george graeme gross hand hearst hinrich homograph http human improved information integration international introduction issue japanese jean jiri journal kanehisa katherine kilgarriff knowledge krogh language large latter learning lexical lexicography line linguistics local lrec makoto manning manual marti method miller minor minoru mitaku mitchison models molecular more nagao nancy natural noun nucleic numerical parser patterns performance present press probabilistic problem proceedings processing programs project proteins references research resolve resources richard satoshi schutze sean sekine sense senseval sentential sequence sequences serious shigeki similarity special state statistical stetina suitable syntactic text that this tion university used using value veronis volume waterloo which with word wordnet would http://acl.ldc.upenn.edu/P/P03/P03-1064.pdf 64 A SNoW based Supertagger with Application to NP Chunking abney academic adjoining algorithms almost based brill case chen chunking chunks cient coling collins computational delaware disambiguation discriminative driven eacl editors emnlp error experiments formal grammars grammatical handbook hidden improving information joshi kluwer kudo language languages learning lexicalized linguistics machines markov matsumoto methods models naacl natural pages parsing part parts perceptron principle proceedings processing publishers references rozenberg salomaa schabes shanker speech springer srinivas statistical study super supertag supertags support tagging theory thesis towards training transformation tree university using vector vijay volume with http://acl.ldc.upenn.edu/P/P03/P03-1061.pdf 61 Morphological Analysis of a Large Spontaneous Speech Corpus in Japanese about accuracy accurately achieved aided alone analysis another approach argamon arti based berger better black both categories cial classi coling committee computational conclusion context corpora corpus could dagan decisiontree della described describes detect detected detecting dictionary distributional editors emnlp engelson entropy estimation eubank extraction formalism found from furui includes information intelligence isahara japanese jaynes kashioka koiso language large levine linguistics lrec maekawa maximum mechanics method methods model mori morphological morphology nagao nagata natural nitions nlprs nobata only page pages paper part physical pietra press probabilistic problem proceedings processing references research review sample second segments sekine selection semiautomatically several speech spontaneous stand statistical than that their theory there this tribus type uchimoto unknown used using when where without word words yamada http://acl.ldc.upenn.edu/P/P03/P03-2019.pdf 90 Integrating Information Extraction and Automatic Hyperlinking adaptation after agents allen annotated asahara automatic based been both bottom building case charles chinese clicking coling compiler complex computational constraint converting corpora covered creation czech dark davis deliverable demonstrator description dfki disambiguation easily entity extended extralink feature figure finite fkhui flexible generated geneva german grammars hand hfnhu high hyperlinks hypertext icon implemented inflection information intelligent interpreted into involves issco java karolinum krafft krieger language languages left lightness linguistic linguistically link links linux lisbon lissabon lrec mapping marked matsumoto mayfield methods mmorph models morphology morphosyntactic multext multilingual named nicholas ontology operational other page palmas part performance petitpierre piskorski piskruvnl polish prague presentation press proceedings processing program przepi references regular report research resources result rich right rkowski runs russell segmentation shallow shows spain speech speed sprout state structures study subramanian tagger tagging target this tools type typed unbearable under unification university window windows with wolinski word workshop http://acl.ldc.upenn.edu/P/P03/P03-1001.pdf 1 Offline Strategies for Online Question Answering: Answering Questions Before They Are Asked aaai abdessamad advances andrew annual answering association banko based berland brill charniak classification coling college computational conference corpora daniel data disambiguation dumais echihabi eduard entities eric eugene exploitation finding fine fleischman france gaithersburg grained hermjakob hovy intensive international jimmy language large linguistics marcu maryland matthew meeting michael michele michelle named natural nist park part parts proc proceedings question references reformulation resource retrieval rule scaling seattle some speech susan tagging taipei taiwan text toulouse trec very washington http://acl.ldc.upenn.edu/P/P03/P03-1053.pdf 53 A Word-Order Database for Testing Computational Models of Language Acquisition acquisition adverbs algorithm application associates based bertolo berwick bias binding briscoe broihier cambridge center characterizing chomsky cinque coevolution cognitive conditions control cuny defined device domain dordrecht erlbaum fifth fodor foris from functional general gibson gold government graduate grammatical heads identification inductive information inquiry journal langley language lawrence learn learnability learners learning lectures limit linguistic mahwah mathematics meeting melnikova minimalist models niyogi oxford parametric parsing press proceedings program proposed psycholinguistic publications recently references report research results science shafto society structurally superparsing syntax systems technical testing triggers troseth unambiguous university wexler http://acl.ldc.upenn.edu/P/P03/P03-1048.pdf 48 Evaluation challenges in large-scale document summarization advances agreement algorithms aligning among analysis assessing automated automatic beverly bilingual bloedorn brandow carletta cation chin church classi computational condensation conference content corpora crochemore daniel differences document documents donna editors eduard electronic eric gale harman hills hovy inderjeet information introduction jean kappa karl kenneth klaus krippendorff linguistics lisa management mani marcu mark maxime maybury methodology mitze orleans oxford pages press proceedings processing program publications references related retrieval rytter sage selection sentence sentences september similarities statistic summarist summarization summarizing tasks text understanding university william wojciech http://acl.ldc.upenn.edu/P/P03/P03-2015.pdf 86 A spoken dialogue interface for TV operations based on data collected by using WOZ method action agent analysis available broadcast captioning cdrom channel communication conference controllers cselt decoder dialogue digital direct education educational eiji electronic evaluation expo facts fipa goto haruo hideki hiruma http human icassp ichiro ieee imaging imai interactive interface international ishihara isono japanese kazuteru komine makino morita multimedia name news nobuyuki noriyoshi opera operations package pass proceeding proceedings program progressive projects realtime receivers references remote search selection services sharon speech spie spoken station sumiyoshi symposium system takao takayuki tatsuya technologies television tion toru toshiya tsuda turn uratani usability using utterances vision volume watch work wowow yagi yamada http://acl.ldc.upenn.edu/P/P03/P03-2006.pdf 77 Finding non-local dependencies: beyond pattern matching algorithm annotated antal antecedents available based beatrice beyond bosch briscoe buchholz building carroll clark coling collins computational conference copestake corpus coverage daelemans deep dependency downloads driven eisner empty english evaluation exploration from graham grammatical guide head hockenmaier http international jakub jason john johnson jonathan julia kaufmann language large learner learning linguistics lrec machine marcinkiewicz marcus mark mary matching meeting memory michael models morgan natural nding nodes pages papers parser parseval parsing pattern penn pennsylvania probabilistic proceedings programs publishers quinlan recovering reference references relation relational report ross sabine santorini schemes simple sloot statistical steedman stephen structures technical their thesis three tilburg timbl treebank university using version walter wide workshop zavrel http://acl.ldc.upenn.edu/P/P03/P03-1020.pdf 20 tRuEcasIng accent accents across additional addressed agreement also although ambiguity analysis andrew anlp another application applications applying articles automatic badly based beyond bikel bleu brill capitalized capture case cased channel chieu church classi comma computational conclusions consider considered context contextual correction cover current david decision decisions decoding depending detection different dimensional direct disambiguation discontinuous discrimination discussed distance distribution during edit enhancing entity error evaluation expressive extended extension feature features followed form forms free french further future gale general genres golding have helpful hidden high icml impact improved improvements improving include information issues item items jing jones kenneth kishore knowledge language latent learning legibility lexical linguistics lists machine markov martin medium method might mikheev miller model modeling models moore more most must name named nder needed news nition noisy normalization notion nymble outside over pages papineni performance performs perhaps phonetic pilot possible potentially process professionally proposed provide punctuation rabiner range readings recognition references report research resolution restoration restoring roth roukos salim schwartz selected semantic sensitive shallow similar since sources spaces spanish speech spelling statistical study styles surface take task teaching text through todd tracking translation truecaser truecasing tutorial upper useful using valu very ward weaker weischedel well which william winnow with word words work would written yarowsky http://acl.ldc.upenn.edu/P/P03/P03-1015.pdf 15 Combining Deep and Shallow Approaches in Parsing German abney adam antonio approach based berger berwick black briscoe carol carroll chunks comparing computation computational coverage darpa della dordrecht editors english entropy evaluation flickinger gdaniec grammars granada grishman grove harrison hindle ingria jelinek john klavans kluwer language liberman linguistics lippo lrec march marcus maximum natural paci pages parser parsing pietra principle procedure proceedings processing proposal psycholinguistics quantitatively references robert roukos santorini speech stephen steven strzalkowski survey syntactic tenny vincent workshop http://acl.ldc.upenn.edu/P/P03/P03-2035.pdf 106 Deverbal Compound Noun Analysis Based on Lexical Conceptual Structure academic adding adjectival adjective aiso ambiguity analysis another approach argu argument association assumptions available based because beth between both breadth busa case cecile coling combining compact comparing complex compositional compound compounds computational computing conceptual conclusion consider constituents dealing department depth designing deverbal deviate dictionary different disambiguate disambiguation discussion domain each ecitons edition editor effective english example experiments fabre factor fails federica found from function future generative grimshaw hale have hideo hirosato iida independent indicate information internal interpretation isabelle jackendoff james jane japan japanese johnston judith justify kageyama kentaro keyser kluwer kurosio language lapata lauer learners levi levin lexical lexicon lexicons liguistics like look looks macquarie malka maria mark meanings michael middle misunderstood model modi most multiple needed nominal nominalization nominals nomura notes noun nouns number ogura ohmusha pages paper papers pierre press principled proceedings processing project proposes publishers pustejovsky rappaport reason references relation relations results right role roles roughly samuel semantic semantics small society speaking speci statistical struc structure structures syntax taro technical terms that them thematic then thesis this typical university verb viegas view what which while wilkins with word words work working http://acl.ldc.upenn.edu/P/P03/P03-1041.pdf 41 Effective Phrase Translation Extraction from Alignment Models alex alignment american annual application arti association automatic based bleu bracketing brown christoph cial coling comparison computational conference copenhagen corpora curin daniel david decoding dekai della denmark discriminative divergence empirical entropy estimation evaluation fast final foster france franz george germany grammars hermann ijcai improved intelligence international inversion jahr john joint josef kenji kevin kishore knight lafferty language linguisics linguistics machine marcu mathematics maximum meeting melamed mercer method methods michael minimum model models montreal natural nist noah north onaizan papeneni parallel parameter peter philadelphia phrase pietra probability proc proceedings processing purdy references report research robert roukos saarbrucken salim segmentation smith speech spoken statistical stephan stephen stochastic summer syntax tests tillmann todd training transduction translation version vincent vogel waibel wang ward william with wong word workshop yamada yarowsky yeyi http://acl.ldc.upenn.edu/P/P03/P03-2005.pdf 76 An Adaptive Approach to Collecting Multimodal Input adaptation adaptive alex also amanda annual anurag applications applied approach architecture association australia australian automobile average bangalore based beijing bernard bers best bliss building cambridge carnegie carpenter china choen choi collected collection compared compositional computational computer computing conclusion conference data deangeli declarative delay demonstrate designing determination determining dialog dialogue dialogues directions duncan during dynamic earlier effectiveness ehlen enhanced environment eric errors experiments factors feature ferro focus framework from further future gesture gunaranjan gupta harry holzman human improved improvement improving increased input inputs integration interaction interactivity interface international interpretation interpreting johnston kuhn landay language larson lauderale learn learning linguistics logic make maloor marilyn match mechanism meeting melbourne mellon method michael minh mmif modal model modelling modes module modules multi multimodal natural number observed online oviatt partly patrick performance phil philadelphia pittsburgh predicting prediction preetam preferred press previous proceedings processing proposed raymond reduced reducing reference references repeated report represents research respectively response results robert robustness same science secs semantic sharon shum speech srinivas state stent steve structures style synchronization system systems tasks technical technology temporal terry tested tests that their this thompson time times turn typed university usability user users using vasireddy vergo waibel wait walker whittaker widows will winograd with work http://acl.ldc.upenn.edu/P/P03/P03-1027.pdf 27 Recognizing Expressions of Commonsense Psychology in English Text aaai about acknowledgments advances aise allen annual applications army associates authors baker bartsch behavioral berk buckland burges cambridge canada children cognitive coling collaboration commerce commonsense competency computer computers conclusions conference contract cooperative coverage daad data department developed development dialogue dicale discours dyer eleventh erlbaum expressed fall fast ferguson fillmore findings formal framenet fran frank from funds gordon hobbs human humanities implementations indexing information institute intex java jorand kaufman kernel knowledge lawrence learning lkopf lowe machine machines mahwah manager march meeting memory menlo mental methods mind minimal mining montreal morgan national necessarily notes opinions optimization oxford pages paper park part plan platt practical practice press proceedings project psychiatrie psychologie reasoning reboul recommendations reconciling references reflect report representations research retrieval revue sabatier schizophr science sciences sequential shatz silberztein smola social society source spring standards stanford state states storybooks strategy support symp symposium synthesizing systems talk technical techniques technology text theories theory this those tools training trains traum trec tude under university using vector views voorhees washington wellman with witten working york young http://acl.ldc.upenn.edu/P/P03/P03-2027.pdf 98 Dialog Navigator : A Spoken Dialog Q-A System based on Large Text Knowledge Base analysis answering base based biatov bocchieri coling commun communicator computational conf conjunctive darpa detection dialog dialogue eckert engine european eurospeech fabbrizio fuyuko icslp japanese julius kawahara kido kiyota knowledge kurohashi language large levin linguistics long makoto method mixedinitiative nagao narayanan navigator open pages pieraccini pokrovsky proceedings processing question rahim real recognition references ruscitti sadao sentences shikano source speech spoken structures syntactic system tech text time vocabulary walker yoji http://acl.ldc.upenn.edu/P/P03/P03-2003.pdf 74 On the Applicability of Global Index Grammars adapting algorithm applicability association barbara becker berlin bloomington bounded castano ciaa cicling cient city communications compilation computing context coverage dassow descriptive earley editors extension february formal free gazdar gigs gils global grammars guages heidelberg hpsg index indexed indiana journal june language languages lopez machinery mathematics mexico natural oehrle parsing polynomial power proc proceeding references regulated restricted rewrita reyle rogers rohrer santa scriptive sensitive springer theory time wide york http://acl.ldc.upenn.edu/P/P03/P03-1035.pdf 35 Improved Source-Channel Models for Chinese Word Segmentation1 achievements algorithm algorithms analysis approach assp based better breaking cambridge chang channel cheng chiang chien chilin chinese christopher class cognitive coling component compression computational conclusion contextual contributions cooling corpus correlated data detection entity estimation extraction factoid feng finite first fold formula formulate from gale gilbert goodman gwyneth huang identification ieee improved including incorporating information integral jasis jerome jian jianfeng joshua katz keyword khoo kowk language linguistics mcnad methods ming mingjing model modeling morphological morphology named nancy ning packard paper phonology preliminary present press probabilities problem problems recognizer references retrieval richard rocling rodger second segmentation shih shing sigir simultaneously solved source sparse speech sproat state statistical stochastic study talip teahan tech text these this three toward tree tseng tung unified university unknown using which william witten wong word yingying yong yubin zhang zhou zimin http://acl.ldc.upenn.edu/P/P03/P03-1068.pdf 68 Towards a Resource for Lexical Semantics: A Large German Corpus with Extensive Semantic Annotation acknowledgments address agreement ambiguity annotated annotation annotations annotator automatic both bottlenecks construct corpus data design designed discussed domain encouraging evaluated extraction focus focused following frame from gures have helped important independent individual information inter lack language large lexica lexical like many merging most next order paper people preferences presented problem project proposed purposes reliably representation results roles salsa scheme selectional semantic semantics semi serious show steps sugges tags technology thank their this today tool underspeci used utilizable utilize vagueness which while with would http://acl.ldc.upenn.edu/P/P03/P03-1014.pdf 14 Integrated Shallow and Deep Parsing: TopP meets HPSG account accuracy acquisition address also analysis annotation arbeitspapiere architecture arti astray automated based baseline bears becker beijing berlin beyond brants briscoe budapest callmeier canada cant cantly carroll case cases chances chunk cial ciency cient clausal coling cologne compared components conclusion constraint constraints could coverage crossvalidation crysmann data daum debusmann deep demonstrated dence dependency derived design different duchier eacl easily editor edmonton emnlp engineering erroneous esslli eurospeech experimentation extended extracting extraction factor factors fail felder figure fine formance foth found foundations fragments france frank from further gain gains german grained grammar grammatical greece grover guided high hohle hong hpsg implementation improve improves induction information infrastruc integrated integration intelligence isolated issues iwpt kasper kiefer kong krieger language lascarides lead length level lexical linear linguistic longer lopar loss losses macrostructural manuscript margin matters menzel msec muller naacl natural noord noun outliers over pages parse parser parses parsing part performance phrases pisko pittsburgh platform postagging potential precedence precision preparation prevented prins prioritisation prioritised proceedings processing range recognition reduces references relations relative report reported results rhodes robust robustness rski sample scenario schafer schmid sealts selection sentence sentences shallow show shows siegel signi significant similar skut some spans speech springer spurious statistical stochastic stuttgart substantial suggests surface syntax system tagger taipei taiwan techniques terms that this time took topological topologische toulouse translation trees ture university unpublished unsupervised uszkoreit variance verbmobil wahlster weights were what where which while with workshop xslt xxvii http://acl.ldc.upenn.edu/P/P03/P03-1016.pdf 16 Synonymous Collocation Extraction Using Translation Information aaai academic acquisition across agustini algorithm annual answering answers artificial association atlanta automatic barzilay base based blair bokyung boston brown carolyn categorization chinese cikm clustering coling computational conference construction contexts corpora corpus correspondence crouch della development dialog ding discovery dragomir eacl ebmt esslli estimating estimation evaluation experiments exploits explorations extracting fifteenth finding from fung gamallo gasperin generation goldensohn grefenstette groups hong huang improving independent information intelligence international issue john kido kiyota kluwer knight knowledge koehn kurohashi langkilde language large lexicalized lima linguistics lopes machine management mathematics mckeown measuring meeting mercer mining model monolingual national natural navigator noisy normalization pacific parallel parameter paraphrases paraphrasing pietra prager press probabilities proc processing question questions radev references research resources retrieval sasha selection shimohata sigir similar similarity sixth special statistical sumita symposium syntactic system taiwan target technical tenth term text that thesaurus third trained translation unrelated using vera waiguo wang with word words workshop yang zhang zheng zhiping zhou http://acl.ldc.upenn.edu/P/P03/P03-1040.pdf 40 Feature-Rich Statistical Translation of Noun Phrases access across additional address algorithm algorithms alignment allows almost also always approach automatic base based being berger best bilingual bleu brown candidates china chunking coling collins comparison complexity compound computational computed conclusions conll consider context data decoder decoding demonstrated discriminative does eacl emnlp empirical enables entities entropy estimation evaluation experiments external fast features foundations framework free full generation generative german germann grammar graphs have hongkong improved instead integrated into introduce jahr knight koehn language languages lexicalized limited linguistics machine malouf manning manual many marcu mathematics maximum mercer method methods models monolingual naacl named natural noun onaizan only optimal over pages pair papineni parameter parsing partial phrase pietra press probabilistic problem proceedings processing reduced references require reranking resources robust roukos schmidt schulte schutze search separated show shown splitting statis statistical subtask system that this three tical training translated translating translation using usually walde ward where which with word yamada http://acl.ldc.upenn.edu/P/P03/P03-2026.pdf 97 Automatic error detection in the Japanese learners’ English spoken data adding although based chunking conference corpora correct data decision decreased found helmut information international jaynes lance language large learning marcus mechanics methods mitchell part physical precision probabilistic proceedings processing ramshaw rate recall references review schmid sentences speech statistical tagging text that theory third through training transformation trees using very went workshop http://acl.ldc.upenn.edu/P/P03/P03-1017.pdf 17 Constructing Semantic Space Models from Parsed Corpora academic accessibility accurate acquisition activation adam allan american analysis anlp associates automatic balota behavior british bullinaria burgess burnard canada choi clustering cognition cognitive cogsci coincidence colin coling computation computational computers computing comrie consortium constraints contextual corpora corpus correction curt david decision dekang depth determinants direct discovery discrimination distributional dumais dunning edinburgh editor effects effort elements emnlp environmental erlbaum experimental explorations extracting francisco freddy from grammar grefenstette gregory guide hastings highdimensional hinrich hodgson induction information informational inquiry instruments james johanna john jones joseph journal kaufmann keenan kevin kilgarriff klincksieck kluwer knowledge landauer language large latat latent lawrence learning levy lexical lillian linguistic linguistics livesay london lorch lowe lrec lucien lund malti martin mcdonald meaning measures mediated memory methods michael model montr moore morgan national neural noun occurrence oxford pages paris patel peter phrase platos prelexical priming problem proceedings processes processing producing pronunciation psychological psychology publishers references representation representations research retrieval review robert route salton science scott seattle segmentation semantic sense service similar similarity society solution space spaces spelling spreading statistics structurale surprise syntaxe tesni text theory thesaurus thesis tools towards universal university users using vector wang what wiemer will word words workshop yallop yang http://acl.ldc.upenn.edu/P/P03/P03-1062.pdf 62 Learning to predict pitch accents and prosodic boundaries in Dutch abney accent acoustical addison albert algorithms analysis annual arti assigning assignment association automatic aviation based belmont berck black bosch boundaries breaks breiman brunswick busser buttersworth caley cambridge cation cial cient classi collins combining communications compared comparison comprehension computational computer conference consistency context corpora cover cstr cutler daelemans dagan dahan data december discovery discrimination discriminatory documentation donselaar edinburgh edition editors ejerhed electrical electronics emnlp engineers english eurospeech evaluation exceptions experts feature fifth forgetting fourth friedman from generator gillis group guide harmful hart herwijnen hirschberg hodges hong hoste human icassp icslp improving information informativeness instance institute intelligence international intonational issue japan jensen john journal kibler king knowledge koehn kohavi kong ladd language large learner learning library linguistics literature local london lrec machine marsi mckeown medicine meeting memory methods mining modeling natural nearest neighbor nonparametric oates ohlsen pages part pattern phonology phrase phrasing pitch placement predicting prediction press proc proceedings processing progressive project prominence properties pros prosodic prosody provost reading reasoning reference references regression report resources retrieval review reynaert rijsbergen salton sampling scandinavia school selection sequences sigdat sloot society sources special speech spoken spring stan stone structure subset syntactic synthesis system tagger tasks taylor technical terken text theory third tilburg timbl tools toward transactions transformation trees university usaf version very wadsworth waltz wang wesley with word workshop wrappers zavrel http://acl.ldc.upenn.edu/P/P03/P03-1033.pdf 33 Flexible Guidance Generation using User Model in Spoken Dialogue Systems aaai acoust adapting adaptive applied arise arti automatic bennacef burianek carberry carroll cecile cial coling collaborative computational concept conf conference consultation dence descriptions dialogue dialogues diane eckert elzer esther evaluation expertise finin flexible gauvain hazen icslp ieee information initiative integrating intelligence interactive jean jennifer joseph kass kawahara kazunori komatani lamel language level levin limsi linguistics litman lori management measures mimic mixed modeling national natural object output over pages paris paul pieraccini polifroni poor predicting preferences proc process processing queries recognition recognizer recognizing report retrieval robert roberto rosset samir sandra scoring seneff shimei signal sophie speech spoken stephanie system systems tailoring tatsuya text theresa timothy track train travel trec understanding user using utilizing wieland with workshop http://acl.ldc.upenn.edu/P/P03/P03-1046.pdf 46 Parsing with generative models of predicate-argument structure acquiring address adwait aggressive algorithm also annual argument arise baseline beam best bounded building cambridge cantly captured captures categorial charniak christoph clark cleaner collins combinatory compact conclusion conditional conference consistent contrast corresponding current czech data deep demonstrate dependencies dependency derivations different driven edinburgh emnlp entropy estimating eugene experiments first from future generative grammar grammars hajic head here hockenmaier implementation important improved including indicate informatics inspired julia kind lance language less lexicalized local long longrange lrec manner mark mass maximum meeting method michael might model models more naacl natural ofspeech overcome overly pages palmas paper parser parsing part particular pennsylvania performance philadelphia predicate presented press probability problems proceedings process proposed question ramshaw range ranker ratnaparkhi references restrict results school search seattle signi simple simpler sound statistical steedman stephen strategy structure structures suitable syntactic tagger that these thesis third this tillmann treebank unbounded university used uses using very whether which widecoverage will with word work http://acl.ldc.upenn.edu/P/P03/P03-2039.pdf 110 Chinese Unknown Word Identification Using Character-based Tagging and Chunking akira analysis application asahara automatic based cation changning chasen chen cheng chinese chunking coling colips computational conference dayang documents extraction hirano hiroshi huang identi implementation institute international jiann kazuma kevin kitauchi kudo language linguistics local machines manual maosong masayuki matsuda matsumoto morphological naacl nara pages ping proceedings processing references regcognition roles science shen sighan statistics support system tagging takaoka taku tatsuo technology unknown vector version with word words workshop yamashita yoshitaka yuji zhang http://acl.ldc.upenn.edu/P/P03/P03-1039.pdf 39 Chunk-based Statistical Translation above accumulated addition alignment alignments allowed alpha alshawi also always amount appearing applied approximated approximation associated bangalore boundary brown chunk chunked chunking chunkings chunks climbing collections combination computation computational computations compute considering corresponding count counts della demand dependency derived determine determines douglas each enumeration estimation even except first following follows formulas forward from given handle have head hill hiyan index initial inside introduction learning length linguistics massive maximum model models move neighboring nite number occurrences operations operators order outside pair pairing pairs paradigm parameter paring particular peter pietra position posj possible probabilities probability procedure random ranging references reorder reordering represent requires respectively second sentence sentences shona sible source srinivas state stephen such swap target through transducers translation translations using values variables vincent viterbi where which with word words yields http://acl.ldc.upenn.edu/P/P03/P03-1065.pdf 65 An Expert Lexicon Approach to Identifying English Phrasal Verbs about academic acknowledgment afrl allows also american anlp anonymous approach architecture associated associates authors automatic automaton baldwin based benchmarking bern between bolinger bond both breidt budapest called cambridge canada carrie church cicling city coling combination comlex comments comparison compiled comprehending computational computer conclusion conducted conference constructions constructive contains contract copenhagen copestake cornell corpus cuny definition description develop directorate discovery dissertation driven dryrun editors edmonton effective efficiency efficient engine engineering english entire erlbaum experiment expert experts expressions extend extraction faster file finite flickinger following force formal formalism francke fraser from future grammar grammars grant handles harvard head hillsdale idarex identification idioms index indexed information infoxtract inseparable institute interaction international intex into korea laboratory language large lawrence lehnert levels lexemes lexicography lexicon linguistics local macros mass massachusetts matching mexico module morpho multi multiword naacl natural neck ninth noun only other over pain papers parser parsing particle partly parts pattern pelli pentium phrasal phrase pine plan platform presented press problem proceeding proceedings processing program provides realisation recognition references regular required requires research reviewers reviewing rieger ringle rome rules runner running same sampling satisfactory science sciullo second seconds segond seoul separable sequential shaked sharon shows silberztein small software solution south srihari state states stochastic strategies structure successful supported supporting syntactic system systems takes task technology test testing text thank thanks that their theoretical theory this times toolbox transitions treatment types under university unrestricted used using valetto verb verbs verlag villavicencio volume walter williams windows wish with word words work workshop york zurich http://acl.ldc.upenn.edu/P/P03/P03-2002.pdf 73 An Ontology-based Semantic Tagger for IE system abney academic amstrong ananiadou anlp answering applications applied approach arti automatic based beijing bengio berkeley boufaden brill california cambridge canada canadian chapter church cial class coast coef combining cone conference context corpora cream design developed dialog dictionaries dictionary disambiguating disambiguation doctorat domain edition encies engineering england enrich european evaluation extraction fifteen fisheries foundations france fridman from future given government granada groupings guard hafner hong http hwee information intelligence international isabelle isbn italy kaufman kluwer kong language lapalme large learning lesk london lrec lyon machine magazine manning manual manzi massachusetts maynard message morgan most natural nlprs noun november oceans ones ontology overlapping pages paper parsing part partial pine preliminaries press proceedings processing public publishing question readable reference references relevant rescue resnik resource respect rule schutze seamanship search segmentation selecting semantic semantically sense senses senseval services several shiuan shriberg sigdoc simple spain speci speech stage state statistical stevenson such summarization symposium tagger tagging techniques tell term that theory thesaurus these third this topic toronto trento tutorial tzoukermann understanding university uses using very with word wordnet words work works workshop yarowsky http://acl.ldc.upenn.edu/P/P03/P03-1002.pdf 2 Using Predicate-Argument Structures for Information Extraction aaai accuracy achieves acquisition adding alternations american annotation annual appelt approach argument arti assigning association automatic automatically baker based bear because berkeley beth better branching brill brown california cambridge canada cascaded case central charles chicago chinchor cial classes classi coling collin collins computational conclusion conference constituents customizable daniel data david decisions degradation diego domain domains douglas driven easily ellen english eric erica error event extracting extraction fast fastus ferro fillmore finite finitestate florian framenet from general generated generating generative germany gildea grace grishman guidelines harabagiu hirschman hobbs html http human huttunen identi identifying inductive info information infrastructure intelligence international investigation israel jerry john jones jurafsky kameyama kingsbury knowledge labeling lane language learning less levin lexical lexicalized linguistics lisa lowe lynette mabry madrid marcus mark martha measure meeting megumi method michael mihai mining mitch models montreal naacl nancy national natural necessity ngai north novel open over pages palmer paper paradigm parsing part pasi patricia patterns paul penn performance philadelphia piece predicate predicates preliminary press problem proceedings processing project proposed quinlan radu ralph recognition references reports riloff robinson roles roman rosie ross rulequest saarbrucken sanda scales semantic silja spain speech state statistical stickel structures study sundheim surdeanu tagging tapainen task technology templettes text than thirteenth this three tools transducer transformation transformationbased treebank tyson university unknown untagged used verb when yangarber http://acl.ldc.upenn.edu/P/P03/P03-1012.pdf 12 A Probability Model to Improve Word Alignment accurate adam aligning alon alshawi amta annotation approach august automatic bangalore berger bilingual brown carbonell change chinese christian church cohesion coincidence collections computational corpora correspondence correspondences darpa della dependency douglas dunning emnlp entropy erik estimation evaluating gale head heidi high hiyan identifying jaime jason katharina kaufmann kolak language lavie learning levin limited linguistics lori machine march mathematics maximum mercer methods models monson more morgan natural nite okan pages parallel parameter peterson philip phrasal pietra precision probst proceeding proceedings processing projection ralf rebecca references resnik resource rule shona small speech srinivas state statistical statistics stephen surprise texts transducers translation translational using vincent weinberg with word words workshop http://acl.ldc.upenn.edu/P/P03/P03-1045.pdf 45 k-valued Non-Associative Lambek Categorial Grammars are not Learnable from Strings aarts alternations amsterdam angluin antecedent appendix applies apply arithmetical aspects associative auxiliary balt based benthem between bonato buszkowski calculus california case cases categorial cation chapter christian clas computa computational computer conference context control counts croisic dana data derivations description detail determined dropping editors either elsevier empty exponents fact finally forgetting formal france francois free from gentzen gerald gives glyn grammar grammars groote handbook have hence hillel holland hypothesis ieee induction inductive inference information inserting insertion instead international introduction june kaufmann lacl lamarche lambek language languages last learning lemma linguistic linguistics logic logica logical mateo mathematical mati meulen middle minimalist morgan morill nbal nbalt north notation obtained only pages penn pentus philippe polynomial positive powerful pregroups presentation press proof quasi quaterly references relationship replacement result retore rewritten rigid roberto rule science second sentences september sequences shinohara show sical similar similarly society springer step strasbourg struc studia style such suppose syntactic that their then theory therefore third three time tional trautwein tured verlag volume where with without wojciech workshop http://acl.ldc.upenn.edu/P/P03/P03-2028.pdf 99 Spoken Interactive ODQA System: SPIQA acquire activated adding addition adopting although analysis answering answers applied approach appropriate arti audio augmented automatic based cial clause coling comparison computer computing conclusion consultations conversational copes disambiguation domain elsevier erman errors eurasip experimental extracting formalism future generate generated grammars hall handling harabagiu harpy hearsay helped holland hori ieee indispensable information integrating intelligence interactive interface journal jupiter knowledge lacking language lowerre lter medical meeting missing mycin networks nite north notes ntcir odqa once only open pereira performed potential prentice processing proposed queries question questions recognition references research resolve results revealed sasaki screening shortliffe signal speech spiqa spoken statistical summarization survays survey system systems taipei technique telephone that third transactions transition trends uncertainty understanding users using voice weather will with working workshop york http://acl.ldc.upenn.edu/P/P03/P03-1010.pdf 10 Reliable Measures for Aligning Japanese-English News Articles and Sentences about adwait akira aligning alignment alignments allowed also anlp approach approximations article articles association automatic availability based been bilingual both boundaries brill church coling collier community company comparison computational contains corpora data development dictionary discussed distribute distributing educational effective electronic encoded english entropy eric from gale generously gengo half have hideki high highschool hirakawa hiroshi houkoku html http identifying ikeda included index industry information inside interested japan japanese jeffrey jeida junior kenji kenneth kumano language late linguist linguistics lrec machine makoto many masaya matching matsumoto maximum members model mutiyama nagao name natural news newspaper nigel nlprelated numerically obvious october organizations other outside over pages paper part people poisson probabilistic processing program psycho purposes ratnaparkhi recover references related report request requests research results retrieval reynar robertson rule school sent sentence sentences sets shimbun sigir simple sizen some speech statistics suru syori system systems tagger takehito tanaka teachers term text that their them these this translation tyousa users using utsuro walker weighted were william yamane yomiuri yuji http://acl.ldc.upenn.edu/P/P03/P03-1019.pdf 19 A Comparative Study on Reordering Constraints in Statistical Machine Translation addisonwesley algorithm algorithms alignment annual apparatus application applications approach aprendizaje april arti association athens august automatic based berger berlin bilingual bleu boostrap bracketing brown catalan center character chinese cial cocke combinatorische complexity computational computer conf constraints context corpora cruz dass december decoding della discrete discriminative division dominio edition editor empirical empleo entropy evaluation examples fast flug foundations fteen fundamental generating generation germany gillett grammars graphs greece grunschnabel hannover hanover hong hotel human ight ijcai improved instance intelligence international into introduced inversion jelinek joint journal july june kehler kings knight knuth kong lafferty language leusch linguistics lrec machine mathematics mathematik maximum meeting mercer method methods models montreal mwer nach natural november number numbers october pages papineni para parallel parsing past patent percola philadelphia physik pietra politecnica polynomial problem probleme proc processing programming quarter reading reference references related report research resources restringido results roossin roukos santa schlage schroder second segmentation september sequences seven shapiro siam sieben source speech springer sser states statistical stay stephens stochastic subsecuenciales suggest suggestion system table tareas task technical that thesis thomas time tion tool training transduction transductores translation translations trees type ubernachten united universidad using valencia verbmobil verlag vier viertel vilar volume vorschlagen wahlster ward watson were west with word wordreplacement work would wurde zeitschrift http://acl.ldc.upenn.edu/P/P03/P03-2031.pdf 102 Automatic Acquisition of Named Entity Tagged Corpus from World Wide Web according acquire agichtein almost also andrew annual association automatic automatically based borthwick boundary brunswick case categories category chang choong chunk chunking coling collect compound computational conclusions conducted conference contain contextual corpora corpus cost could demonstrated detection diverse document documents does dongseok engine enormous enough entities entity entropy equally estimation eugene examine exceeds experiment experimentally exploiting fewer focused from gary generates generation germany gets geunbae grishman guodong however hybrid hyeok hyun improved increase increased information instances integrated international internet into japan japanese jeong jeongwon jersey jeung jian john jong journal jungyun junhyeok knowledge korean language large learning less lexical linguistics list machine manabu management manual maximum meeting merit method methods million more morpheme much multistrategic nally named natural nement nitely nouns nyoung only outperforms paci pages paper partof pattern patternselection performance person philadelphia point posed preference presented procedures proceedings processing ralph recognition references result rule rules same sassano saturation science search searched segmentation segmented segmenting sentence sentences seok seon separation shim shown sixth size society sources speech statistical sterling suggested supervised syllable symposium system table tagged tagger tagging takehito techniques text than that therefore this through tokyo trained unknown useful using utsuro very when which whole with without word words workshop youngjoong zhou http://acl.ldc.upenn.edu/P/P03/P03-2008.pdf 79 A Ranking Model of Proximal and Structural Text Retrieval Based on Region Algebra algebra alin alon buckley burkowski charles chinenyanga chris cient clarke collection computer cormack daniela data databases david deutsch evaluation expressive fernandez florescu forbes framework fuhr gordon hersh hickam http implementation information interactive journal kototoi kushmerick language large leone levy mary models nicholas norbert ohsumed probabilistic proceedings query querying ranked references research retrieval search sigmod structured suciu taurai test text webdb william workshop http://acl.ldc.upenn.edu/P/P03/P03-1058.pdf 58 Exploiting Parallel Texts for Word Sense Disambiguation: An Empirical Study adam annual association brown chugur classifier combination computational conference consensus corpora cotton david della dependence description diab directions disambiguation discrimination domain edmonds empirical english erjavec escudero evaluating felisa florian future gerard german gonzalo international irina joint julio kilgarriff language large lexical linguistics lluis marquez meeting mercer method methods modeling mona nancy natural overview pages parallel peter philip pietra polysemy proceedings processing proximity radu recent references resnik rigau robert sample scott second sense senseval sigdat siglex statistical stephen study successes suite supervised systems tagging task test tomaz tufis unsupervised using verdejo very vincent with word wordsense workshop yarowsky http://acl.ldc.upenn.edu/P/P03/P03-1069.pdf 69 Probabilistic Text Structuring: Experiments with Sentence Ordering acknowledgments acoustics alex alistair also american annotated approach aravind arti asher association author available barbara barzilay based beatrice bottom building cambridge canada centering chapter charniak chris cial clark cohen coherence columbia combining comments component computational computer conditional conference conversation corpus dale daniel data dekang dependency discourse donnell editors ehud eleazar elhadad english entropy eskin estimation eugene evaluation experiments framework francisco frank from fusion generation geoffrey global granada grant grateful grosz hatzivassiloglou helpful hoffmann ieee inferring information inspired integrated intelligence international island israel john joshi journal judith karamanis kathleen katja katz kaufmann keller kibble klavans knott lafferty language large lascarides learning lebanon linguistics local logics lrec machine making marcinkiewicz marcu marcus markert mary maximum mckeown mellish mick miles minipar mitchell mitzpe model modeling models morgan multidocument national natural news nicholas nikiforos noemie north number oberlander ontario order ordering orlando osborne oxford pages parser parsing penn permutations planning power praphrasing press probabilities probability proceedings processing progress pronominalisation prospects providence providing publishers ramon rankings recognizer references reformulation regina reiter research rhode richard robert rodger sammut sampson santorini schapire scott search seattle sentence signal singer slava software sparse speech stephen stochastic strategies suggestions summarization supported systems text thanks their thesis things this towards transactions treebank university using valuable vasileios weinstein william work workshop yoram http://acl.ldc.upenn.edu/P/P03/P03-2022.pdf 93 An Evaluation Method of Words Tendency using Decision Tree El-Sayed Atlam, Masaki Oono, and Jun-ichi Aoe Department of Information Science and Intelligent Systems University of Tokushima Tokushima,770-8506, Japan. E-mail: atlam@is.tokushima-u.ac.jp ABSTRACT analyzing articles artificial automated automatic based cartoon character classification clustering communication computational conference contents contracts data decision definitions demand dictionary discourse electronics engineering english european evaluation extracting extraction field format from fukumoto generating gonick groups guide hara haraguchi harpercollins haruo hill hisano honda human ieice importance important indexing information inoue institute intelligence interest introduction ipsj japan japanese journal kaufmann keyword keywords kitani language learning liman linguistics litman login machine mcgill mcgraw means mochizuki modern morgan nakajima natural newspaper observation occurrence ohkubo okumura page passonneau phrase proceeding processing programs publishers quinlan rate references reflect research retrieval sagara salton search segmentation series signl smith society some special specific statistics sugizaki summarization suzuki tanaka terms text that time trans tree trees type unbalanced using variations watanabe word york http://acl.ldc.upenn.edu/P/P03/P03-1018.pdf 18 Orthogonal Negation in Vector Spaces for Mo delling Word-Meanings and Do cument Retrieval academic accessing acquisition adapted addison american analysis annals annual april aspects associates automatic available baeza because berthier between birkho bloomington boolean both branches bruce buckley calculated cambridge canada chris christiane ciency cient collection combining communications composing computation computational conclusions concrete conference connectives considerable contribute could crimination croft database david demonstration developed developing discrimination disjunction document documents dominic dumais dunlop each ective editor edmonton edward electronic evaluation example experiments exploited extended feedback fellbaum fertility focussing form from gain gains garrett gerald gerard harry have hersh hickam hinrich hope howard http implementation improving indiana individual inference infomap information inherent initial interactive introduction involve involving john journal just kluwer kowalski landauer language large latent learning leone lexical linguistics logic ltering mark mathematics mcgill mcgrawhill meaning meanings mechanics methods michael modern multiple naacl natural negation neto networks neumann nonmatching norwell novemb ohsumed over pages performance peters plato post potential powerful preliminary press problem proceedings processing product psychological publicly publishers quantum queries query references related relevance remove removing representing research retrieval retrieved review ribiero ricardo rigourous salton scalar schutze science semantic sense shown sigir similarity society solution some spaces stanford stanley statement statistical structure susan syntactic system systems tasks taxonomies techniques terms test that their theory these this thomas tool traditional transactions turtle unique unsupervised unwanted used using vector vectors webdemo wesley which widdows will william word wordnet yates york http://acl.ldc.upenn.edu/P/P03/P03-1055.pdf 55 Deep Syntactic Processing by Combining Shallow Methods adam american annual approach association berger bies bracketting budapest building cambridge chapter charniak christoph cient clark college collins computational conference consortium coverage czech data deep della dependency entropy eugene european ferguson generative guidelines hajic hockenmaier implementation inspired julia karen katz lance language learning lexicalised linguistic linguistics macintyre madrid mark maryland maximum meeting michael model models natural north park parser parsing penn philadelphia pietra press proceedings processing project ramshaw references rens robert seattle statistical steedman stephen structures style three tillmann treebank university vincent wide with http://acl.ldc.upenn.edu/P/P03/P03-2036.pdf 107 Comparison between CFG filtering techniques for LTAG and HPSG abeille acknowledgment acquisition adjoining advantage algorithm also annotated anonymous another application approximated approximation becker between building cfgs chicago cient coling comments communications companion compared comparison computational concluded conclusion conduct constraints context control conversion converting corner corpus degree different direction driven eacl earley effective empirical engineering english equivalent essential esslli establish existing experimental experiments extracted february fellowships free from functional future given global going grammar grammars group harbusch head help helpful hpsg information integrate intelligence interface into investigating ircs iwpt joshi jsps kaplan kiefer krieger language languages large lexicalized ling linguistics ltag ltags lter ltering makino marcinkiewicz marcus maxwell miyao more natural ninomiya nishida noord obtain obtained offs order pacling pages parallelization parser parsers parsing partially penn pennsylvania phrasal phrase pollard poller presented press proc produced recognition references report research results reviewers revisited sakao santorini schabes scientists session showed size step strategies strongly structure student style supported technical than thank that their this time torisawa trade tree treebank tsujii university using volume ways will with work xtag yoshida yoshinaga young younger yousuke http://acl.ldc.upenn.edu/P/P03/P03-2032.pdf 103 Extraction and Verification of KO-OU Expressions from Large Corpora analysis annual association beforehand brings certainly chew chiga concord corpus elements estimation expression following forth grammar html http inai ipsj isahara iwanami jourmal kakari kamisimereba kanzaki kida kitto kurosiosyuppan lamguage many masuoka measure meeting members modality musubi mutiyama natural notes ohno proceedings processing projects references relation relationship research sake salmon shoten similarity software suru syntax taste umemura which will yamamoto http://acl.ldc.upenn.edu/P/P03/P03-1021.pdf 21 Minimum Error Rate Training in Statistical Machine Translation advances algorithms alignment analysis annual arpa automatic ayuso bangalore bayes berlin bilingual byrne cation chou classi coding conference cruz discriminative doddington duda editors empirical evaluation generation george germany goodman gram hart human international israel john joshua juang june kumar language lopez machine meeting methods metrics minimum mitzpe natural occurrence pages parsing pattern peter philadelphia proc proceedings processing quality rambox ramon recognition references richhard risk rubio santa scene shankar soler speech springer srinivas statistical statistics technology texts translation trends using verlag whittaker wiley william workshop york http://acl.ldc.upenn.edu/P/P03/P03-1009.pdf 9 Clustering Polysemic Subcategorization Frame Distributions Semantically according accurate acquisition allaria allerton alternation alternations analysis annotation annual applied argument automatic balanced based behaviour between bialek boguraev bottleneck brew briscoe building business cambridge canada canaria carroll carter cation chicago class classes classi clustering cognition coling comlex communication comparing computational computing conference conll construction contemporary control copenhagen corpora cues dang database denmark derivation dictionary differences directions disambiguation discovery distributions document dorr edmonton elder empirical english evaluation extensions extraction feature features foreign frequency from future gahl general german germany grammaticallyindexed gran grishman grover improving inducing information interlingual international intersective investigating jackendoff japan joanis jones journal jurafsky kipper klavans korhonen kyoto language large learnability levin lexical lexicography lexicon line linguistics longman machine macleod massachusetts master menn merlo method methods meyers miller montreal multilingual natural news noisy pages palmas palmer paradigm pennsylvania pereira philadelphia philadephia pinker predicting preiss press proc processing purely recent references regular resources riddoch robust roland role rosenzweig saarbrucken scale schulte semantic semantically semantics semi sense space spectral stanford statistical stevenson structure structures subcatecorization subcategorisation subcategorization successes supervised syntactic syntax text their thesis tishby toronto translation tsang tutoring university using verb verbs walde washington with word wordnet workshop http://acl.ldc.upenn.edu/P/P03/P03-1057.pdf 57 Feedback Cleaning of Machine Translation Rules Using Automatic Evaluation acquired adam akiba algorithm alignment application arul automatic automatically based best bilingual boundary chart coling conference constituent corpora distances edit eiichiro example extraction from furuse grishman harmonized hierarchical hitoshi iida imamura issues kenji knowledge kosaka language machine mappings menezes methodological meyers michiko multiple natural nlprs osamu output paci pages parsing pattern phrase proceedings processing ralph rank references richardson rule stephen sumita summit symposium theoretical transfer translation using viii with workshop yasuhiro http://acl.ldc.upenn.edu/P/P03/P03-1071.pdf 71 Discourse Segmentation of Multi-Party Conversation acoustic advances analysis appear attention baron beeferman berger biometrika characteristics choi cochran coling columbia comparison computational corpus cues direction disambiguation discourse domain edwards ellis empirical expository galley gelbart giving grosz hajime hearst hirschberg hong html http icassp icsi icslp independent indicators intentions intonational janin kong lafferty learning linear linguistic linguistics litman machine manabu matched meeting models monologues morgan multi multiple naacl nakatani pages paragraph percentages peskin pfau phrases proc prosodic references research samples segmentation segments shriberg sidner some statistical stolcke structure studies surface takeo text topic with wooters http://acl.ldc.upenn.edu/P/P03/P03-2040.pdf 111 TotalRecall: A Bilingual Concordance for Computer Assisted Translation and Language Learning aclclp adaptive align aligning alignment analyses annual assisted association automatic based berkeley bilingual canada chang chinese chuang church cognates collocation companion computational computer conclusion concordance corpora corpus criteria cross dekai describe designed distribution durand english extraction foster free gale information isabelle langlais language length lexical linguistics lrec macklovitch meeting memory mexico mining montreal paper parallel preparing proc proceedings program punctuations references retrieval sentence sentences sigir simard statistical statistically syntactic texts this translation transsearch using wide with world http://acl.ldc.upenn.edu/P/P03/P03-1003.pdf 3 A Noisy-Channel Approach to Question Answering adam andrew answering applied austin banko berger berkeley brill brown charles church clarke computational conference cormack curin data david della dumais eric estimation final franz gaithersburg gordon information intensive jahr jimmy john josef kenneth kevin knight lafferty language linguistics lynam machine mathematics mclearn melamed mercer michael michele multitext natural nist noah noun onaizan parameter parser parts peter phrase pietra proceedings processing program purdy question references reinforced report retrieval robert second sigir smith statistical stephen stochastic summer susan text thomas translation trec unrestricted vincent workshop yarowsky yaser http://acl.ldc.upenn.edu/P/P03/P03-1038.pdf 38 Self-Organizing Markov Models and Their Application to Part-of-Speech Tagging annual applications applied association brants chichester collection computation computational conference emnlp empirical entropy estimating hidden http icopost ieee informatik ingo language learning linguistics logic markov maximum meeting memory methods model models nats natural papers part proceedings processing rabiner ratnaparkhi recognition references schr selected singer speech statistical symposium tagger taggers tagging tbilisi theory topologies tutorial unihamburg using vapnik variable wiley http://acl.ldc.upenn.edu/P/P03/P03-2034.pdf 105 A speech interface for open-domain question-answering acceptable access accessing accuracy acknowledgements allan answerbus answering approach arpa audio authors automatic auzanne balasubramanian barnett based cambridge chang chen cient clarkson colineau comp components computer conclusion conf conference content continue correcting correction crestani data demo described designed devices diego document eacl engineering entry errors esca eurospeech evaluation experiments from fung future garofolo given greece halber helen human hybrid ieee indicates information input integration intelligent interaction interactive interface interfaces kimber knowledge kubin kupiec lang language lect like ltering meng methods misrecognitions mobile modal modeling models moral multi multimedia multimodal necessary nist notes occurrence overview pages palm paper perspectives phones plainsboro pocket port probably proc processing publication queries query question questions rapid recognition references retrieval rhodes rosenfeld ruger scale scho search seide semantic should small speakers special speech spoken statistical stefan story success suggestions summarization support symp system tech text thank that this tien tightening toolkit track trans transcribed trec uncorrected used using voorhees which will with workshop would ying zheng http://acl.ldc.upenn.edu/P/P03/P03-2033.pdf 104 A Debug Tool for Practical Grammar Development absorbed adjoining akane alep always annotation applications applied aravind assumed automatically available axel based because bender bigger biocomputing biomedical callmeier cannot cation cause causes clues coling complicated computational conclusion conference consider content controll conversion corpora correspondent cult debug debugged debugger debugging declerck defects deletes detmar developed developing development difference distributed document emily engineering english enough environment environments errors esslli evaluation event expand extraction failed failure features flickinger formalisms from full future global gotz grammar grammars graphical group handle hasida have hence hisao hpsg http human ichi icts imai implementation infer information inserting ircs january japan japanese joshi kazuma kentaro koiti label labels language large lean lexicalized lidis linguistic longest ltag matsumoto melanie meurers miyao miyata modify module moves naoki natural notes oepen outputs paci pages papers parallel paroubek parsed parser parsing partial patrick paul pennsylvania perspective phase platform point practical preprocess problem proc processing provides recording reduce references remains report research resolution resolve resolved results rieder schabes schmidt semi september session should sibylle siegel simple sixth society stephan student subsentences succeeded successfully support symposium system table tagged tags takaoka takashi tateisi technical that theo theory there these thierry thilo tool torisawa translating tree tsujii university upenn uses using volume walt which will willex with words work workbench workload workshop xtag yakushiji yoshinaga yuji yuka yusuke yves http://acl.ldc.upenn.edu/P/P03/P03-1004.pdf 4 Fast Methods for Kernel-based Text Analysis advances algorithm analyisis analysis anlp appendix application array based binary cascaded case cation chris chunking cient ciently classi coef coling collins conference conll convolution corpus craig cristianini data dependency digital discovery double duffy emnlp engineering entity expanded follows forest frequent given growth hideki hideto hisashi huma icml ieee information international isozaki japanese jian jiawei john journal junichi kashima kazawa kernels knowledge koyanagi kudo kurohashi kyoto lance language learning lemma lodhi machine machines makoto marcus matsumoto michael mining mitchell mohammed naacl nagao nakagawa named natural nature nello neural nigel nips note number otherwise pages part patterns proc proceedings processing product project projected proof ramshaw recognition references research revision sadao same saunders search semi sequential shawe size software speech springer statistical string structure structured subsets support systems tagging taku taylor teruo tetsuji text that theory this thus transactions transformation trees university using vapnik vector vladimir watkins where with xspan yuji zaki http://acl.ldc.upenn.edu/P/P03/P03-1029.pdf 29 An Improved Extraction Pattern Representation Model for Automatic IE Pattern Acquisition analyzer analyzing appos arikawa arimura arrest asai based case conference data databases dependency discovery european hiroki hiromi html http information japanese juman kawasoe kenji kiyotaka knowledge kurohashi language makoto masaki message morphological murata mushoku nagao natural optimized ozaku parser person pkdd post practice principles proceedings qing references resource resources retrieval sadao semi setsuo sharable shinji sixth structure structured substructure system tatsuya uchimoto understanding utokyo workshop http://acl.ldc.upenn.edu/P/P03/P03-1044.pdf 44 Counter-Training in Discovery of Semantic Patterns aaai accuracy acquisition among annl annual applicable applied automatic automatically belonging blum bootstrapping cambridge categories categorized cation cited classi coling college collins colt combat combine combining compared competition computational concept conclusion conf contexts convergence copenhagen counter data degradation dependency dictionaries disambiguation domain each effective emnlp empirical entity exactly expansion experiments extraction features from generalized generating good grishman have huttunen information intl jarvinen joint jones kind knowledge labeled language learner learners learning level lexicons linguistics literature main meeting method methods mitchell models more multi mutual named names natl natural orlando other otherwise overall paper parallel park parser pattern patterns points precision presented prevent proc process processing projective recall references reports results riloff rivaling saarbrucken search self semantic sense settings several show sigdat simple singer single spotter strategies strengthening strzalkowski supervised supervision taipei tapanainen target termination text that thelen theory they this tradeoff trained training universal unlabeled unlimited unsupervised untagged using wang washington where with word yangarber yarowsky york http://acl.ldc.upenn.edu/P/P03/P03-1063.pdf 63 Text Chunking by Combining Hand-Crafted Rules and Memory-Based Learning accuracy arti based bosch case cation cherkassky chunking cial classi combining computational concepts conll cover daelemans data exceptions forgetting from golding guide harmful hart http ieee improving information intelligence john language lcgwww learner learning machine memory methods mulier natural nearest neighbor pattern reasoning reference references rosenbloom rule shared sloot sons task theory tilburg timbl transactions university version wiley zavrel http://acl.ldc.upenn.edu/P/P03/P03-2024.pdf 95 A Limited-Domain English to Japanese Medical Speech Translator Built Using R E G U L U S 2 able aching adjectives after allow alone annual april association being bouillon budapest burning cambridge carter cation causal cause character compiling computational conference constrained conventions corresponding demo desu different digalakis doctor does dowding dull each eacl editor editors eighteenth elliptical elliptically english enough environment event example february feel found foundations from generated germany given grammar grammars grammatically hageshii have here hockey however html http hungary important instead integratedwavetechnologies interaction international into investor james japanese language like line linguistically linguistics major maps medical meeting models more motivated much natural needs nuance obvious okoru only open pain philadelphia phrase phrasebook phrased possible postpos press prob problem proceedings projects question questions rapidly rayner recognisers references regulus relates representation saarbrucken samuelsson second semantically sentence series some source sourceforge specialisation speech spoken springer stand style target that there these thus track transfer translation translator tsukare typed types university used utterances uzukuyona verbmobil very wahlster what which wiren http://acl.ldc.upenn.edu/P/P03/P03-2013.pdf 84 Approaches to Zero Adnominal Recognition adjunct adnominal advances aizawa akio akira amsterdam analysis anaphora anaphoric annual approach approaches aravind aspects association automatic barbara based bean benjamins between bond botley case centering classification cognitive coherence cohesion coling complement complex computational conference connected copula corpus corpusbased david definite definiteness description descriptions determination determiners dictionarybased discourse dissertation distinction editors ellen empirically english expressions fais forgotten framework francis fujiwara grammar grosz group halliday harold hasan heine hideo hiromi hirosato hituzi identification ikehara independence indirect information interest international into investigation iori isao issues iwanami japan japanese john joshi julia kawaoka kentaro kentarou kojima kurohashi kurosio language laurel lexicon linguistic linguistics links local longman machine makoto masahiro masaki massimo mathematical mcenerry meaning meeting methodological miho mitsuko miyazaki modeling murata nagao naito nakaiwa natural nominal nomura noun nouns number ogura osaka oyama pacific particle particles philadelphia phrase phrases poesio possessive prediction proceedings processing pronouns property publishers publishing quebec ranking reference references referential relations renata report review riloff ruqaiya sachiko sakahara sakai salience satoru satoshi scott semantic sentences shigeru shimazu shirai shozo society somers special stanford structure structures syntax syobo system taikei taipei takei teachers technical teramura teruaki texts theoretical toenglish tokyo translation tsukasa under understanding university unpublished valency validity vieira weinstein with yamura yokoo york yoshie yoshifumi zeros http://acl.ldc.upenn.edu/P/P03/P03-1034.pdf 34 Integrating Discourse Markers into a Pipelined Natural Language Generation Architecture adjoining aggregation analysis anchoring annual anyway appeared aravind architecture august background based bonnie canada carla chris christine circsim clause coling columbia complex comprehensive conference consensus david december delaware discourse donia doran ehud embedded english european evens explanations extended feng fifth france from generating generation glass grammar greenbaum high historical human interaction international jacques james joint joshi journal jung just kennebunkport knowledge lake language leech level lexical lexicalized lexicon linguistic longman manfred markers martha meeting mellish memory michael mike montreal mooney natural newark niagara pages plausible power proceedings providing psycholinguistically publishers quirk reape references reiter relations revision rhetorical richard robin schemata schrock scott seventh shaw speech spontaneous stede structure summaries svartvik systems text thesis toulouse tree tutor tutoring umbach understanding university urbana usage using variable webber what workshop yang http://acl.ldc.upenn.edu/P/P03/P03-1050.pdf 50 Unsupervised Learning of Arabic Stemming using a Parallel Corpus acknowledgements across advanced agency alexander algorithm aligned also analysis annual appear arabic association ballesteros based brown buckwalter clark companion computational conference connell contract corpora daniel david defense della diab emam estimation free goldsmith grace hany hassan helpful hidden html http identifying improving inducing induction info information john july jurafsky kishore knowledge language languages larkey latent leah learning light like linguistics lisa lisbon machine margaret markov master mathematics matthew meeting mercer method misspellings model models mona monitored morphology multilingual natural ngai observations occurrence ossama pages pair papineni parallel parameter partially patrick philip pietra proceedings projection projects references research resnik retrieval reviewers richard robust roukos salim schone segmentation semantic sense sigir snover spawar spon stemming supported tagging text thank their thesis this tools translation translitchart transliteration under university unsupervised upenn using volume washington wicentowski with word work would yarowsky young http://acl.ldc.upenn.edu/P/P03/P03-2021.pdf 92 iNeATS: Interactive Multi-Document Summarization accurate anlp automatic behavior belkin bikel british canada carbonell case coincidence columbia computational computing conference daniel development documents dunning edmundson effectivness evaluation extraction factors goldstein highperformance human information interaction interactive jade jaime journal jurgen kantrowitz koenemann learning linguistics mark methods metrics miller mittal name nder nicholas nymble pages proceedings ralph references research retrieval richard schwartz scott selection sentence sigchi statistics study summarizing surprise systems text vancouver vibhu weischedel http://acl.ldc.upenn.edu/P/P03/P03-1023.pdf 23 Coreference Resolution Using Competition Learning Approach acquisition anaphora anaphoric annual aone approach association automated based bean bennett brennan burger centering chinatsu computational connolly coreference corpus david dennis evaluating extraction friedman identification information john joseph language learning linguistics machine manual massachusetts mccarthy meeting methods noun page pages phrases pollard proceedings processing pronouns reference references resolution riloff scott strategies thesis trainable university http://acl.ldc.upenn.edu/P/P03/P03-1013.pdf 13 Probabilistic Parsing for German using Sister-Head Dependencies aaai about achieves alternative american anette annotated annotation annual applied arti association atness based baseline beatrice becker been beil beyond bikel both brants brigitte building cambridge canaria carroll categorial chapter charniak chiang chinese christoph cial college collins combinatory computational conclusions conference considered constituent context corpora corpus csli curves czech daniel data david dependencies design detlef ects empirical english entropy estimation eugene european evaluation evidence experiment explanation failed frank franz free from full generative german gildea glenn grammar gramotron gran granada hajic half hans head helmut hockenmaier hong however hypothesis implementation improved induction inside inspired institute intelligence international investigated julia kong krenn lack lance language languages large learning lexical lexicalised lexicalized linguistics literature lopar lrec madrid makes marcinkiewicz marcus mark markus mary mats maximum measures meeting methods michael mitchell model models montreal national natural nding negra north odds order outside palmas park parser parseval parsing part partial particular particularly pcfg penn perform performance philadelphia pittsburgh poor possible precision prescher presented press probabilistic proceedings processing produce properties publications ramshaw recall references reported results riezler rooth sabine santorini scheme schmid schulte seattle semi show sisterhead size skut sparse speech standard stanford statistical statistics steedman stefan stochastic structure structures stuttgart substantially suffer suitable syntactic syntactically systems tagger taipei text that therefore they this thorsten three tillmann topological towards trained training treebank university unlexicalized unrestricted uses uszkoreit valence variation very walde washington what which with wojciech word wordorder workshop worse http://acl.ldc.upenn.edu/P/P03/P03-2041.pdf 112 Learning Non-Isomorphic Tree Mappings for Machine Translation aaai abeille ability able above absent accommodated according accuracy actual acyclic address adjoining adjunction advantage after again agenda algorithm algorithms align aligned aligning alignment alignments allow allowed allows almost alshawi also alternates although analogous analogy analyzers analyzes annotated appear appeared appears apply approach arbitrary arises assigns associate author available avoid await baiser balance bangalore based beaucoup because been before best between biggest bilingual binary bohmova both bottom bound bounded branching break brings brown building builds call cannot carefully case cfgs channel chart child children choice choose chooses chosen cient clever clsp coling collected collections combines comp complex complexity complication compression computational compute condition conditional connected considers consists constant constitute constructing contain context contiguous contributed converges corpora corpus correspond correspondence corresponding could count counts create cross csli culty cycle czech daniel data december decoder decoding decomposed decomposing decorate deep dekai delete deletion denote dependency derivation derivations derived derives describe desired determined developed developing different directed discourage discuss discussed distinct distribution does donnent douglas during each edge edges editors effort elementary elle else emnlp empty english enough enriching ensure entirely enumerate equals equivalent essentially estimate estimates exactly example except exception exible exibly expands expectation expected explicitly exponentially expository expressed extending extract fast features finally following forced forest form formalism formally former forms formulate formulated fortunately found four free freely freetranslation french from frontier full fully function further generalization generalize generate generates generation given gives global gloss goodman grammar grammars grammatical grant graph greatly guarantee gure hajic hajicova handle handled hash have having head hence here hladka hopkins however hypotheses hypothesize identity idiomatic ignore implementation implemented impose improved improves improving include includes including incoming increment indeed indexed indicate initial initially inner innovations input insert insertion inside instead interactions interested interesting internal into inversion irene isomorphic iterates itself johns joint jointly joshua just keep kenji kevin khalil kids kiss kluwer knight labeled labels langkilde language languages last latter learn learning least leaves leaving left less like limited limiting line linear ling linguistically linguistics literature local long lookup loop lots machine mainly many marcu match matching matchings maximization maximum meaningful means mechanically members mentioned method methods might minimal mismatched mismatches miss model modeling models more morpheme most much must naacl naive nalement nally natural need needed nite nition node nodes nonterminals note noted null number numbers observation obtains obvious offers often ones only operation optimal order oriented other others outer outgoing output outside over pair paired pairing pairs paper paragraph parallel parameters parse parsed parsing part partie past pcfgreductions perhaps peripheral permit perspective phrase phrases picked plugging polynomial poor possible post potential potentially poutsma powerful prague preferences prepared present presented presents previous prior priori probabilistic probabilities probability probable problem problems proc procedure proceedings process processing product proportion provide provided prune putting quantities quite randomly rather real reasonable reasonably reasons recorded recover recursive reestimate reestimating regarded related relied remain remaining remko rens replace replaced replacement replaces require required requires researchers resembles restrict restriction result results return rich rigidity roles root rooted roughly rule rules running runtime said salad same scfg scfgs scha schabes second section selectional semiring sense sentence sentences serve several shallow shieber should shrink sibling side sima similar simple simplistic since single sister sity size sketched slower small solely some somehow source space spans speaking speci special specify stag standard star state states statistical statistics status step steps still stochastic stored strategies stray strictly string structure stsg stsgs stuart style subgraph subgraphs subject subset substitute substitutes substituting subtree subtrees such summarization summarized summer supervision supported surface symbol synchronous syntactic syntactically syntax target team tendency terminal terms text than that their them then there these they third this thus till tillmann time times track trainable trained training transducers transduction transform translate translation translations tree treebank treebanks trees tried tsgs tting tuple unaligned unary unbounded unboundedly uncertain under unfortunately units univer unknown unless unlike unmatched unpaired upper used useful uses using usually value values variety version versions vertex views viterbi ways weakly well were what when which whose wider will with within without word words work workshop would write yamada yielding yves http://acl.ldc.upenn.edu/P/P03/P03-1022.pdf 22 A Machine Learning Approach to Pronoun Resolution in Spoken Dialogue aberdeen abstract acts allen alon amsterdam analysis anaphor anaphora annual antecedent applications applied approach approaches april arti asher association atkinson automatic based belmont bonnie breiman briscoe brooks bruckstein burger byron cardie carroll cation centering charles christoph chung cial claire classi cognitive cole college colloquium computational computer conference connolly coreference corpora corpus daarc dagan daniel data deixis demonstrative dennis dialogue discourse distance distributed donna dordrecht eckert edit eiichiro elizabeth elsevier empirical entities evaluation extraction feature feldman foundation friedman from gentleman george graphical graphics hirschman hwee ihaka improving intelligence interpretation introduction itai james jerome joel john journal july kaufmann kazuhide kluwer kohavi language learning linguistics lter lynette machine marc march mateo mayo meeting meng message methods michael minimum miriam model morgan muller natural netherlands nicholas noun objects olshen ostension package park partitioning paul penn philadelphia phrases plying preference proc proceedings processes processing pronominal pronoun pronouns rapp recursive reference references regression report resolution resolving robert ross routines rpart scheme scoring second selection semantics soon statistical statistics stefan stone strube structure subcategorization subset sumita synchronising technical terry tetrault theoretic therneau towards training trains trees uence understanding units using vilain vincent vision wadsworth washington webber with workshop wrappers yamamoto yong http://acl.ldc.upenn.edu/P/P03/P03-2009.pdf 80 Discourse chunking: a tool in dialogue act tagging aaai accept action acts after alexandersson allen annotation appeared backchannel before beginning buschbeck cambridge chunk chunking chunks clarify classifiable classified close coding comment commit communicative confirm core corpus correctly damsl defer deliberate deviate dialogs dialogue discourse examples exclude explained fall feedback first formula four fujinami give greet group humans inform information init machines maier measure more negative notes okay percent politeness positive precision reason recall references reithinger reject report represents request results right scenario scheme schmitz second shows siegel signal some suggest symposium table than thank that those times types utterances verbmobil with without wolf word working yeah