Idiomify -- Building a Collocation-supplemented Reverse Dictionary of English Idioms with Word2Vec for non-native learners

What it is about

The project builds a reverse dictionary for idioms, so that a learner can find an idiom from a phrase or a definition, and adds collocates of each idiom so the learner can use it precisely. Idioms are taken from the SLIDE list of frequent idioms, and 2,746 are kept after dropping short ones. Idiom matching rules are generated from base forms and cover three of the six variation types: optional hyphen, inflection and alternative wording. Modification, passivisation and open slot fail, although the author shows how slop, wildcard and reordering could cover them. Idioms are found in over 3 million utterances of spoken corpora. Collocates are ranked by term frequency, PMI and TF-IDF; on five idioms with dictionary example sentences as the reference, PMI and TF-IDF rank the collocates better than frequency, with no clear winner between them. The reverse search averages word2vec vectors of the phrase and returns the nearest idioms; its median rank of 128 on a 100-idiom test set is not comparable with the Hill et al. baselines, as the author notes. Removing stopwords makes the median rank worse (128 against 156).

Ideas

• I1. en/src/morph.rs: the matching rules as explicit lemma patterns with alternatives and an optional hyphen, written as data, with one test per variation type. Add slop (a word gap limit) for modification, and a reordered pattern for passives, with a negative control that must not fire on unrelated sentences.
• I2. 19. The lexicon: vocabulary, multiword expressions and register: keep stopwords in the idiom-to-definition index. The ablation in the paper (128 against 156 median rank) is the first measure to repeat with Mova's own idioms.
• I3. Count-based collocates: compute PMI and frequency from counts in a BTreeMap, not a library model, and check the ranking against dictionary example sentences, with a shuffled-count control.
• I4. A hybrid search: an inverted index of definitions first, then a ranking by the phrase vector only when the index is empty. The author proposes this as future work; it is also the explainable order.

For Mova

The chapters are 19. The lexicon: vocabulary, multiword expressions and register (idioms and collocations) and 34. Irony and figurative language (figurative use). The code points are en/src/morph.rs for the matching rules, world/src/dictionary.rs for definitions, and the pupil dictionary in en/src/dict.rs. The measure: the match rate per variation type on a held-out idiom list, and the median rank of the definition search with and without stopwords, run twice for determinism.

Cite

The paper itself, not this card. BibTeX from the library's record of the work. arXiv 2204.05634

@misc{kim2022idiomify,
  title = {{Idiomify -- Building a Collocation-supplemented Reverse Dictionary of English Idioms with Word2Vec for non-native learners}},
  author = {Kim, Eu-Bin},
  year = {2022},
  eprint = {2204.05634},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2204.05634}
}

References

16 references, 1 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Arora, S. Liang, Y. Ma, T. (2017). A Simple but tough-to-beat baseline for sentence embeddings Beeferman, D. (2003). OneLook reverse dictionary. Available at: http://onelook.com/reversedictionary.shtml. ↗
2. Caro, R. Edith, E. (2019). The Advantages and Importance of Learning and Using Idioms in English. find
3. Word Association Norms, Mutual Information, and Lexicography Davies, M. (2008). The Corpus of Contemporary American English (COCA). Available online at https://www.english-corpora.org/coca/. ↗
4. Ganesan, K. (2020). What are Stop Words?. Available at: https://kavita-ganesan.com/what-are-stopwords/#.YKfXMy0Rr_Q ↗
5. Granger, S. (1998). Prefabricated patterns in advanced EFL writing: collocations and formulae. In: Phraseology: Theory, Analysis and Applications. Clarendon Press: Oxford. 145-160. find
6. Gormley, C. Tony, Z. (2015). Elasticsearch: The Definitive Guide; O’Reilly Media. In: Sebastopol, CA, USA. find
7. Hill, F. Cho, K. Korhonen, A. Bengio, Y. (2016). Learning to Understand Phrases by Embedding the Dictionary Honnibal, M. Montani, I., (2017). spaCy 2: Natural language understanding with Bloom embeddings, convolutional neural networks and incremental parsing. find
8. Hughes, C. Filimonov, Maxim. Wray, Alison. Spasi´c, Irena. (2021). Leaving No Stone Unturned: Flexible Retrieval of Idiomatic Expressions from a Large Text Corpus. find
9. Ingebrigsten, M. (2013), Indexing for Beginners, Part 3. Available at: https://www.elastic.co/blog/found-indexing-for-beginners-part3 ↗
10. Jochim, C. et al. (2018). SLIDE: a Sentiment Lexicon of Common Idioms. Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC’2018). European Languages Resources Association (ELRA) find
11. Manning, D., et al. (2008). Tf-idf weighting. Viewed on 23th of February 2021. Available at: https://nlp.stanford.edu/IR-book/html/htmledition/tf-idf-weighting-1.html Miller, G. (1995). WordNet: A Lexical Database for English. Communications of the ACM Vol. 38, No. 11: 39-41. ↗ our card
12. Oxford Advanced Learner’s Dictionary of Current English. (1995). Oxford University Press Oxford Collocations Dictionary for Students of English. (2002). find
13. Oxford University Press Rehurek, R. Sojka, P. (2011). Gensim–python framework for vector space modelling. NLP Centre, Faculty of Informatics, Masaryk University, Brno, Czech Republic, 3(2). find
14. Sierra, G. (2000). The onomasiological dictionary: a gap in lexicography. In Proceedings f the Ninth EURALEX International Tiedmann, J. (2012). Parallel Data, Tools and Interfaces in OPUS. Proceedings of the Eight International Conference on Language Resources and Evaluation (LREC’2012). European Language Resources Association (ELRA) find
15. Thorat, S. Choudhari, V. (2016). Implementing a Reverse Dictionary, based on word definitions, using a Node-Graph Architecture. find
16. Thyrab, R. (2016). The Necessity of idiomatic expressions to English Language learners Yang, S. (2003). machine learning for collocation identification find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/kim-2022-idiomify-building-collocation-supplemented-reverse/