LING848, FALL 2013.

Advanced Seminar in Computational Linguistics: Computational Social Science

Description

Back in 2009, Mark Liberman argued on Language Log that "corpus based social science" was poised to go mainstream, despite a general historical tendency toward "linguistic anemia" in the social sciences. As computational linguists, we have known for the past several decades that language use in corpora can serve as a useful proxy for world knowledge. With so much of people's lives going online, it was time to start imagining technology that would exploit large-scale language use as evidence for people's individual properties, behaviors, and social interactions, as well.

Since then, there's no doubt that Mark was right. A September 9, 2013 Time magazine article entitled "What Twitter Says to Linguists" mentions that "upwards of 150 Twitter-based studies have come out in 2013 so far", and has a nice quote from Jacob Eisenstein, whose work is featured in the article: "Language is really a window into people's sense of personal identity".

Now, "social science" is an unmanageably huge topic. The Wikipedia page on Social Science includes disciplines ranging from archaeology to social work. In this seminar, we'll narrow the field somewhat. I'm particularly interested in the idea of perspective, in the Oxford Dictionary sense of "a particular attitude toward or way of regarding something; a point of view". This seems to me to be a focal issue in the social sciences: our perspectives help to define who we are, and they influence the decisions we make and the nature of the other individuals and groups with which we consider ourselves associated. Within computational linguistics, sentiment analysis is a familiar subset of perspective emphasizing the way that language communicates positive, neutral, and negative polarity, but there are many other ways of looking at perspective that are to my mind equally (and more!) interesting.

With that in mind I'm going to organize the content primarily, although perhaps not exclusively, around two main themes.

• The first of these is computational political science, a research topic is trending strongly right now. There the idea of perspective emerges strongly in discussions of ideology and framing. Some example questions of interest... How can we identify the ideological positions of individuals from the language they use? How is language used to emphasize or deemphasize aspects of an issue under discussion in order to persuade or influence others? How do we computationally characterize and model influence in political discourse?

• The second is computational approaches to psychology. This may seem incredibly different as a domain, and that's ok -- diversity keeps things interesting -- but at the same time I think there are some important commonalities even though here computational modeling approaches are just beginning to emerge: psychological properties of individuals (e.g. gender, neuroticism) and clinical conditions (such as depression) are closely connected with the way that those individuals perceive the world and how they respond to it. Some example questions of interest... How can we identify psychological properties or conditions of individuals from the language they use? How can we track progress (or decline) of clinical conditions over time in a minimally invasive or intrusive way?

This seminar will mainly involve readings and in-class discussion, helped along by participation in discussions on Piazza. The class will be graded on participation (30%), which includes leading class discussions, as well as a term paper/project (70%). I hope to encourage hands-on projects that involve real problems, aiming for papers suitable for submission to appropriate conferences.

Philip Resnik, Professor
Department of Linguistics and Institute for Advanced Computer Studies


Department of Linguistics
1401 Marie Mount Hall            UMIACS phone: (301) 405-6760       
University of Maryland           Linguistics phone: (301) 405-8903
College Park, MD 20742 USA	   Fax: (301) 314-2644 / (301) 405-7104
http://umiacs.umd.edu/~resnik	   E-mail: resnik AT umd _DOT.GOES.HERE_ edu