Abstract / Description of output
This paper presents a winning submission to the SemEval 2022 Task 1 on two sub-tasks: reverse dictionary and definition modelling. We leverage a recently proposed unified model with multi-task training. It utilizes data symmetrically and learns to tackle both tracks concurrently. Analysis shows that our system performs consistently on diverse languages, and works the best with sgns embeddings. Yet, char and electra carry intriguing properties. The two tracks' best results are always in differing subsets grouped by linguistic annotations. In this task, the quality of definition generation lags behind, and BLEU scores might be misleading.
Original language | English |
---|---|
Title of host publication | Proceedings of the 16th International Workshop on Semantic Evaluation (SemEval-2022) |
Editors | Guy Emerson, Natalie Schluter, Gabriel Stanovsky, Ritesh Kumar, Alexis Palmer, Nathan Schneider, Siddharth Singh, Shyam Ratan |
Place of Publication | Seattle, United States |
Publisher | Association for Computational Linguistics |
Pages | 75-81 |
Number of pages | 7 |
DOIs | |
Publication status | Published - 1 Jul 2022 |
Event | The 16th International Workshop on Semantic Evaluation 2022 - Duration: 14 Jul 2022 → 15 Jul 2022 Conference number: 16 https://semeval.github.io/SemEval2022/ |
Workshop
Workshop | The 16th International Workshop on Semantic Evaluation 2022 |
---|---|
Abbreviated title | SemEval 2022 |
Period | 14/07/22 → 15/07/22 |
Internet address |