Local contexts4/25/2023 In: Proceedings of 30th Conference on Neural Information Processing Systems, pp. 1006–1017 (2020)ĭai, Z., Yang, Z., Yang, Y., Carbonell, J., Le, Q.V., Salakhutdinov, R.: Transformer-XL: attentive language models beyond a fixed-length context. In: Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pp. 2854–2864 (2020)īlevins, T., Zettlemoyer, L.: Moving down the long tail of word sense disambiguation with gloss informed bi-encoders. 1957–1967 (2017)īevilacqua, M., Navigli, R.: Braking through the 80% glass ceiling raising the state of the art in word sense disambiguation by incorporating knowledge graph information. In: Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pp. In: Proceedings of 7th International Conference on Learning Representations (2019)īastings, J., Titov, I., Aziz, W., Marcheggiani, D., Sima’an, K.: Graph convolutional networks for text classification. 3159–3166 (2019)īaevski, A., Auli, M.: Adaptive input representations for neural language modeling. ![]() In: Proceedings of the Advancement of Artificial Intelligence, pp. KeywordsĪI-Rfou, R., Choe, D., Constant, N., Guo, M., Jones, L.: Character-level language modeling with deeper self-attention. The experimental results by using a series of benchmark WSD datasets show that our method is comparable to the state-of-the-art WSD methods which utilize only the limited number of sense-tagged data, especially we verified that dependency structure and POS features contribute to performance improvement in our model through an ablation test. ![]() By using hidden states obtained by GCNs, Transformer-XL learns local and global contexts simultaneously, where the global context is obtained from a document appearing with the target words. GCNs integrates different features from local contexts, i.e., full dependency structures, words with part-of-speech (POS), word order information into a model. Our model is Transformer-XL framework which is coupled with Graph Convolutional Network (GCNs). This paper focuses the problem and proposes a method for effectively leveraging a variety of contexts into a neural-based WSD model. Supervised Word Sense Disambiguation (WSD) has been one of the popular NLP topics, while how to utilize the limited volume of the sense-tagged data and interpret a diversity of contexts as relevant features remains a challenging research question.
0 Comments
Leave a Reply.AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |