The goal of this annotation is to reconcile different
dependency treebank annotations into a common format. This
involves mapping language-specific part of speech (POS) tags into
generic POS tags, and language-specific dependency relations (RELs)
into generic RELs.
For each POS tag, you have a list of the five most common words in
that tag, plus 20 more, randomly selected from short sentences. You
can mouse-over the examples to view sentences in which they are used.
Your goal is to map the language-specific POS tag to a generic POS
tag. The set of generic POS tags is shown in a draggable window to
the right. You choose the desired POS tag from a drop-down menu as
in:
| Example POS drop-down box: |
Note that if you select "Other", you are allowed to type in an
option. One place where "Other" is necessary is for the English POS
"TO", which conflates both Auxiliary and Adposition uses of the word
"to". When you choose "Other", be sure to fill in the text box with
what you think the tag(s) should be. You have space at the bottom to
add generic comments.
For relations the story is basically the same. There is a list
of valid relations in a draggable box, and you have a drop-down menu
like the following:
| Example REL drop-down box: |
The only added complication is that you need to ensure that the
dependency points in the correct direction. The color-coding
is there to help you. If the dependence direction is wrong, select
the "Flip" checkbox.
Please make notes in the comment section if ANYTHING looks funny to
you (eg., including things like bad preprocessing).
If you are unsure about any annotation, please feel free to contact
me (aka Hal) and let me know!!!
Here are the languages we need annotations for:
|