Until now, there has not been a systematic investigation of the properties of entity linking datasets and their impact on system performance. However, these scores should be considered in relation to the properties of the datasets they are evaluated on. State-of-the-art entity linkers achieve high accuracy scores with probabilistic methods. Proceedings of the 27th International Conference on Computational LinguisticsĪssociation for Computational Linguistics Systematic Study of Long Tail Phenomena in Entity Linking To support this goal, we provide a list of recommended actions for better inclusion of tail cases.", With our findings, we hope to inspire future designs of both entity linking systems and evaluation datasets. We find the most difficult cases of entity linking among the infrequent candidates of ambiguous forms. Our systematic study of these hypotheses shows that evaluation datasets mainly capture head entities and only incidentally cover data from the tail, thus encouraging systems to overfit to popular/frequent and non-ambiguous cases. In this paper we report on a series of hypotheses regarding the long tail phenomena in entity linking datasets, their interaction, and their impact on system performance. Publisher = "Association for Computational Linguistics",Ībstract = "State-of-the-art entity linkers achieve high accuracy scores with probabilistic methods. Cite (Informal): Systematic Study of Long Tail Phenomena in Entity Linking (Ilievski et al., COLING 2018) Copy Citation: BibTeX Markdown MODS XML Endnote More options… PDF: = "Systematic Study of Long Tail Phenomena in Entity Linking",īooktitle = "Proceedings of the 27th International Conference on Computational Linguistics", Association for Computational Linguistics. In Proceedings of the 27th International Conference on Computational Linguistics, pages 664–674, Santa Fe, New Mexico, USA. Systematic Study of Long Tail Phenomena in Entity Linking. Anthology ID: C18-1056 Volume: Proceedings of the 27th International Conference on Computational Linguistics Month: August Year: 2018 Address: Santa Fe, New Mexico, USA Venue: COLING SIG: Publisher: Association for Computational Linguistics Note: Pages: 664–674 Language: URL: DOI: Bibkey: ilievski-etal-2018-systematic Cite (ACL): Filip Ilievski, Piek Vossen, and Stefan Schlobach. To support this goal, we provide a list of recommended actions for better inclusion of tail cases. Abstract State-of-the-art entity linkers achieve high accuracy scores with probabilistic methods.
0 Comments
Leave a Reply. |