Are Word Embeddings Really a Bad Fit for the Estimation of Thematic Fit?

Emmanuele Chersoni; Ludovica Pannitto; Enrico Santus; A. Lenci; Chu-ren Huang

Are Word Embeddings Really a Bad Fit for the Estimation of Thematic Fit?

Emmanuele Chersoni
, Ludovica Pannitto
, Enrico Santus
, A. Lenci
, Chu-ren Huang

Department of Language Science and Technology

Research output: Chapter in book / Conference proceeding › Conference article published in proceeding or book › Academic research › peer-review

6 Citations (Scopus)

Abstract

While neural embeddings represent a popular choice for word representation in a wide variety of NLP tasks, their usage for thematic fit modeling has been limited, as they have been reported to lag behind syntax-based count models. In this paper, we propose a complete evaluation of count models and word embeddings on thematic fit estimation, by taking into account a larger number of parameters
and verb roles and introducing also dependency-based embeddings in the comparison. Our results show a complex scenario, where a determinant factor for the performance seems to be the availability to the model of reliable syntactic information for building the distributional representations of the roles.

Original language	English
Title of host publication	Proceedings of the Twelfth Language Resources and Evaluation Conference
Editors	Nicoletta Calzolari, Frédéric Béchet, Philippe Blache, Khalid Choukri, Christopher Cieri, Thierry Declerck, Sara Goggi, Hitoshi Isahara, Bente Maegaard, Joseph Mariani, Hélène Mazo, Asuncion Moreno, Jan Odijk, Stelios Piperidis
Publisher	European Language Resources Association (ELRA)
Pages	5708-5713
Publication status	Published - May 2020
Event	12th International Conference on Language Resources and Evaluation, LREC 2020 - Marseille, France Duration: 11 May 2020 → 16 May 2020

Conference

Conference	12th International Conference on Language Resources and Evaluation, LREC 2020
Country/Territory	France
City	Marseille
Period	11/05/20 → 16/05/20

More information

Cite this

Chersoni, E., Pannitto, L., Santus, E., Lenci, A., & Huang, C. (2020). Are Word Embeddings Really a Bad Fit for the Estimation of Thematic Fit? In N. Calzolari, F. Béchet, P. Blache, K. Choukri, C. Cieri, T. Declerck, S. Goggi, H. Isahara, B. Maegaard, J. Mariani, H. Mazo, A. Moreno, J. Odijk, & S. Piperidis (Eds.), Proceedings of the Twelfth Language Resources and Evaluation Conference (pp. 5708-5713). European Language Resources Association (ELRA). http://hdl.handle.net/10397/90389

Chersoni, Emmanuele ; Pannitto, Ludovica ; Santus, Enrico et al. / Are Word Embeddings Really a Bad Fit for the Estimation of Thematic Fit?. Proceedings of the Twelfth Language Resources and Evaluation Conference. editor / Nicoletta Calzolari ; Frédéric Béchet ; Philippe Blache ; Khalid Choukri ; Christopher Cieri ; Thierry Declerck ; Sara Goggi ; Hitoshi Isahara ; Bente Maegaard ; Joseph Mariani ; Hélène Mazo ; Asuncion Moreno ; Jan Odijk ; Stelios Piperidis. European Language Resources Association (ELRA), 2020. pp. 5708-5713

@inproceedings{ade244815a754732bef3fddc3a5e4786,

title = "Are Word Embeddings Really a Bad Fit for the Estimation of Thematic Fit?",

abstract = "While neural embeddings represent a popular choice for word representation in a wide variety of NLP tasks, their usage for thematic fit modeling has been limited, as they have been reported to lag behind syntax-based count models. In this paper, we propose a complete evaluation of count models and word embeddings on thematic fit estimation, by taking into account a larger number of parameters and verb roles and introducing also dependency-based embeddings in the comparison. Our results show a complex scenario, where a determinant factor for the performance seems to be the availability to the model of reliable syntactic information for building the distributional representations of the roles.",

author = "Emmanuele Chersoni and Ludovica Pannitto and Enrico Santus and A. Lenci and Chu-ren Huang",

year = "2020",

month = may,

language = "English",

pages = "5708--5713",

editor = "Nicoletta Calzolari and Fr{\'e}d{\'e}ric B{\'e}chet and Philippe Blache and Khalid Choukri and Christopher Cieri and Thierry Declerck and Sara Goggi and Hitoshi Isahara and Bente Maegaard and Joseph Mariani and H{\'e}l{\`e}ne Mazo and Asuncion Moreno and Jan Odijk and Stelios Piperidis",

booktitle = "Proceedings of the Twelfth Language Resources and Evaluation Conference",

publisher = "European Language Resources Association (ELRA)",

note = "12th International Conference on Language Resources and Evaluation, LREC 2020 ; Conference date: 11-05-2020 Through 16-05-2020",

}

Chersoni, E, Pannitto, L, Santus, E, Lenci, A & Huang, C 2020, Are Word Embeddings Really a Bad Fit for the Estimation of Thematic Fit? in N Calzolari, F Béchet, P Blache, K Choukri, C Cieri, T Declerck, S Goggi, H Isahara, B Maegaard, J Mariani, H Mazo, A Moreno, J Odijk & S Piperidis (eds), Proceedings of the Twelfth Language Resources and Evaluation Conference. European Language Resources Association (ELRA), pp. 5708-5713, 12th International Conference on Language Resources and Evaluation, LREC 2020, Marseille, France, 11/05/20. <http://hdl.handle.net/10397/90389>

Are Word Embeddings Really a Bad Fit for the Estimation of Thematic Fit? / Chersoni, Emmanuele; Pannitto, Ludovica; Santus, Enrico et al.
Proceedings of the Twelfth Language Resources and Evaluation Conference. ed. / Nicoletta Calzolari; Frédéric Béchet; Philippe Blache; Khalid Choukri; Christopher Cieri; Thierry Declerck; Sara Goggi; Hitoshi Isahara; Bente Maegaard; Joseph Mariani; Hélène Mazo; Asuncion Moreno; Jan Odijk; Stelios Piperidis. European Language Resources Association (ELRA), 2020. p. 5708-5713.

Research output: Chapter in book / Conference proceeding › Conference article published in proceeding or book › Academic research › peer-review