Hypernym Discovery via a Recurrent Mapping Model

Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021, pages 2912–2921 August 1–6, 2021. ©2021 Association for Computational Linguistics 2912 Hypernym Discovery via a Recurrent Mapping Model Yuhang Bai BDBC and SKLSDE Beihang University, China baiyh@act.buaa.edu.cn Richong Zhang∗ BDBC and SKLSDE Beihang University, China zhangrc@act.buaa.edu.cn Fanshuang Kong BDBC and SKLSDE Beihang University, China kongfs@act.buaa.edu.cn Junfan Chen BDBC and SKLSDE Beihang University, China chenjf@act.buaa.edu.cn Yongyi Mao School of EECS University of Ottawa, Canada ymao@uottawa.ca Abstract Hypernym discovery aims to identify all possi- ble hypernyms of a given term. The most re- cent hypernym discovery models exploit mul- tiple mapping functions to project a term to different semantic spaces and then aggregate these embeddings to a general representation for further classification. We refer to this model as a parallel style model. In this work, we observe that there are hierarchical relations between a target terms’ hypernyms. How- ever, these hierarchical relations were not suf- ficiently considered in the previous parallel style model. To leverage the hierarchical rela- tions, we propose a sequential style model that recurrently maps the query words to their hy- pernyms, starting from the most specific ones to the less specific ones. Empirical studies on SemEval-2018 Task 9 confirm the effective- ness of the presented model. 1 Introduction Hypernymy, namely “is-a” relation, is a vital lexical-semantic relation in natural languages, which relates general terms to their instances or subtypes. In a hypernymy relation, we name a specific instance or subtype hyponym and its re- lated general term hypernym. For instance, (apple, fruit) is in hypernymy relation, where apple is a hyponym and fruit is one of its hypernyms. Due to its general representation ability of semantic re- lations, hypernymy becomes an essential concept in modern natural-language research, and hyper- nymy detection becomes a fundamental component in many natural language processing (NLP) tasks, such as taxonomy construction (Snow et al., 2006; Navigli et al., 2011), semantic search (Hoffart et al., 2014; Roller et al., 2014; Roller and Erk, 2016), textual entailment (Dagan et al., 2013; Bowman ∗Corresponding author Figure 1: The parallel modeling (left) and sequential modeling (right). The Dejan is the name of a basket- ball player. The Bas Player denotes the Basketball Player hypernym. et al., 2015; Yu et al., 2020) and question answer- ing (Yahya et al., 2013; Gupta et al., 2018). One branch of existing works builds the hypernymy-relation-identification problem as a “detection” task, which is only interested in whether a given term pair “is” or “is not” in hypernymy re- lation. These works formulate hypernym detection as a binary classification task. This hypernym de- tection task has been studied for years and plenty of models have been successfully applied in this task (Held and Habash, 2019a; Le et al., 2019). This paper focuses on another problem named hypernym “discovery”, which is different from the detection task. Given an input term, the hyper- nym discovery task retrieves a ranked list of its suitable hypernyms from a large corpus. For train- ing, some hyponyms with their gold …

Content truncated to 500 words