[Corpora-List] Software Release: ADW - Semantic Similarity of Arbitrary Lexical Items

Taher Pilehvar pilehvar at di.uniroma1.it
Mon Oct 13 11:13:03 CEST 2014


* Apologies for cross posting*

--------------------------------------------------------------------------------------------------------------------- Align, Disambiguate, and Walk (ADW) Semantic similarity of arbitrary pairs of lexical items, from word senses to texts! ---------------------------------------------------------------------------------------------------------------------

The Linguistic Computing Laboratory of the Sapienza University of Rome is pleased to announce the first release of ADW.

ADW is a software for measuring semantic similarity of arbitrary pairs of lexical items, from word senses to texts. The software is based on "Align, Disambiguate, and Walk" [1], a WordNet-based state-of-the-art semantic similarity approach presented in ACL 2013.

Features in ADW:

- State-of-the-art performance at multiple lexical levels [1] (word similarity on data-sets such as RG-65 and TOEFL, sentence similarity - the STS task, and WordNet sense clustering). - Cross-level semantic similarity, i.e., between different types of lexical items (e.g., a word sense and a phrase). - Available via easy-to-use Java APIs. - Off-the-shelf: no need for any training or tuning of parameters. - Suitable for upcoming SemEval-2015 Tasks 1, 2, and 3

To obtain the software, please visit ADW's github repository at:

https://github.com/pilehvar/ADW

You can also try ADW's online demo at:

http://lcl.uniroma1.it/adw/

Reference

[1] M. T. Pilehvar, D. Jurgens and R. Navigli. Align, Disambiguate and Walk: A Unified Approach for Measuring Semantic Similarity. Proc. of the 51st Annual Meeting of the Association for Computational Linguistics (ACL 2013), Sofia, Bulgaria, August 4-9, 2013, pp. 1341-1351. -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: text/html Size: 15969 bytes Desc: not available URL: <https://mailman.uib.no/public/corpora/attachments/20141013/6a69441a/attachment.txt>



More information about the Corpora mailing list