Home Research Demos Downloads Publications


Dependency Database The dependency database shown here is extracted from a parsed newspaper corpus. Given a word, one can retrieve from this database the words that had dependency relationship with the input word as well as their frequency counts.
Comparing Word Dependency Relationships This demo takes two words as input. It then align the dependency relationships between each of the word and other words. The related words are shown in three colors. The green words are related to the first word. The blue words are related to the second word. The red words are related to both words. Intuitively, the higher the percentage of the red words the more similar these two input words are.
Dependency-based Word Similarity Given a input word, one can retrieve the words that are most similar to it. The similar words are grouped automatically into clusters. Each cluster is represented by a distinct color. When a word has multiple senses, different clusters often correspond to different senses.

The dependency-based word similarity data can be downloaded here (30MB).

Proximity-based Word Similarity In the above demo, word similarities are computed using the dependency relationships between words. Word similarity can also be computed using the proximity relationship between words.

The proximity-based word similarities can be downloaded here (35MB).

Dependency Index The dependency database only provides the frequency counts of the dependency relationships. Dependency index can be used to retrieve the sentences in which a dependency relationship occurred.
Word Clusters This is a demo of the word clustered obtained with the CBC algorithm.
Paraphrase Discovery This is a demo of the paraphrase discovery algorithm in this paper
Search collocation by Exemplar The other interfaces to the collocation database involves grammatical codes such as 'V:obj:N', 'N:mod:A', ... This interface allows you to search collocations by providing an exemplar phrase.
Usage Checker Second language speakers of English often have difficulty choosing words that are most appropriate among a set of words that are conceptually possible. For example, strictly speaking, perhaps one should be able to say 'cancel a ban'. However, a much more appropriate expression is 'lift a ban'. This program helps you to choose appropriate words.