Detail publikace

Deep Learning from Web-Scale Corpora for Better Dictionary Interfaces

OTRUSINA, L. SMRŽ, P.

Originální název

Deep Learning from Web-Scale Corpora for Better Dictionary Interfaces

Anglický název

Deep Learning from Web-Scale Corpora for Better Dictionary Interfaces

Jazyk

en

Originální abstrakt

This paper explores advanced learning mechanisms - neural networks trained by the Word2Vec method - for predicting word associations. We discuss how the approach can be built into dictionary interfaces to help tip-of-the-tongue searches. We also describe our contribution to the CogALex 2014 shared task. We argue that the reverse response-stimulus word associations chosen for the shared task are only mildly related to the motivation idea of the lexical access support system. The methods employed in our contribution are briefly introduced. We present results of experiments with various parameter settings and show what improvement can be expected if more than one answer is allowed. The paper concludes with a proposal for a new collective effort to assemble real tip-of-the-tongue situation records for future, more-realistic evaluations.

Anglický abstrakt

This paper explores advanced learning mechanisms - neural networks trained by the Word2Vec method - for predicting word associations. We discuss how the approach can be built into dictionary interfaces to help tip-of-the-tongue searches. We also describe our contribution to the CogALex 2014 shared task. We argue that the reverse response-stimulus word associations chosen for the shared task are only mildly related to the motivation idea of the lexical access support system. The methods employed in our contribution are briefly introduced. We present results of experiments with various parameter settings and show what improvement can be expected if more than one answer is allowed. The paper concludes with a proposal for a new collective effort to assemble real tip-of-the-tongue situation records for future, more-realistic evaluations.

Dokumenty

BibTex


@inproceedings{BUT111643,
  author="Lubomír {Otrusina} and Pavel {Smrž}",
  title="Deep Learning from Web-Scale Corpora for Better Dictionary Interfaces",
  annote="This paper explores advanced learning mechanisms - neural networks trained by the
Word2Vec method - for predicting word associations. We discuss how the approach
can be built into dictionary interfaces to help tip-of-the-tongue searches. We
also describe our contribution to the CogALex 2014 shared task. We argue that the
reverse response-stimulus word associations chosen for the shared task are only
mildly related to the motivation idea of the lexical access support system. The
methods employed in our contribution are briefly introduced. We present results
of experiments with various parameter settings and show what improvement can be
expected if more than one answer is allowed. The paper concludes with a proposal
for a new collective effort to assemble real tip-of-the-tongue situation records
for future, more-realistic evaluations.",
  address="Association for Computational Linguistics",
  booktitle="Proceedings of the 4th Workshop on Cognitive Aspects of the Lexicon (CogALex)",
  chapter="111643",
  edition="NEUVEDEN",
  howpublished="online",
  institution="Association for Computational Linguistics",
  year="2014",
  month="july",
  pages="22--30",
  publisher="Association for Computational Linguistics",
  type="conference paper"
}