WordNet: A Lexical Database for English

What it is about

The paper is a short progress report for the WordNet project at Princeton. WordNet groups words into synonym sets (synsets) that stand for lexicalised concepts, and links them with pointers for synonymy, antonymy, hyponymy, meronymy, troponymy and entailment. As of 9 February 1994 it has 81,658 synsets and 109,570 unique strings, and it grows by about 1,500 synsets a month. A semantic concordance tags 103 passages of the Brown Corpus with WordNet senses. The report also describes word-sense identification tests, where local context and WordNet similarity measures improve sense choice, and it lists 280 registered users in 25 countries.

Ideas

• I1. onto: measure the coverage of the open English WordNet, the 120,564 synsets in onto, on EWT test content words. Report the share of lemmas with at least one synset, and the share with none, so the gap is visible next to the 1994 growth figures.
• I2. onto/src/wsd.rs: build a small semantic concordance from the school silver corpus, with one sense per noun token chosen by onto/src/wsd.rs and a hand-checked sample of 100 tokens per passage. Measure the sense accuracy per passage, following the paper's 103-passage design.
• I3. Relation inventory: count which pointer types (synonymy, hyponymy, meronymy, troponymy, entailment) the onto queries and onto/src/wsd.rs actually use, and count how often each one decides a WSD choice on the school silver corpus; a relation that never decides anything is a candidate to drop from the signature.

For Mova

The code is onto/src/wsd.rs and the onto crate, with en/src/expert.rs for the rules that use relation types. The chapter is 19. The lexicon: vocabulary, multiword expressions and register. The paper is an old report with a 1994 lexicon, so use its counts only as a historical reference, not as a target. Measure first: synset coverage of EWT test content words in onto, with two identical runs before any claim.

Topics: Word sense disambiguation

Cite

The paper itself, not this card. BibTeX from the library's record of the work. DOI 10.3115/1075527.1075662

@inproceedings{miller1994wordnet,
  title = {{WordNet: A Lexical Database for English}},
  author = {Miller, George A.},
  booktitle = {Human Language Technology Workshop (Plainsboro, NJ), 1994},
  year = {1994},
  doi = {10.3115/1075527.1075662},
  url = {https://doi.org/10.3115/1075527.1075662}
}

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/lex-1994-miller-wordnet/