Towards Building a Multilingual Sememe Knowledge Base: Predicting Sememes for BabelNet Synsets

Abstract

A sememe is defined as the minimum semantic unit of human languages. Sememe knowledge bases (KBs), which contain words annotated with sememes, have been successfully applied to many NLP tasks. However, existing sememe KBs are built on only a few languages, which hinders their widespread utilization. To address the issue, we propose to build a unified sememe KB for multiple languages based on BabelNet, a multilingual encyclopedic dictionary. We first build a dataset serving as the seed of the multilingual sememe KB. It manually annotates sememes for over 15 thousand synsets (the entries of BabelNet). Then, we present a novel task of automatic sememe prediction for synsets, aiming to expand the seed dataset into a usable KB. We also propose two simple and effective models, which exploit different information of synsets. Finally, we conduct quantitative and qualitative analyses to explore important factors and difficulties in the task. All the source code and data of this work can be obtained on https://github.com/thunlp/BabelNet-Sememe-Prediction.

Automatic Construction of Sememe Knowledge Bases via Dictionaries

“…We can see that the distilled version of EDSKB has overall higher consistency than HowNet, and the full version of EDSKB yields lower consistency results. It is not strange because CCSA is based on sememe prediction and according to previous work (Qi et al, 2020a), senses with more sememes usually have lower prediction performance. Since the full version of EDSKB has much more sememes per sense than HowNet, it is actually not fair to compare their consistency using CCSA.…”

“…Qi et al (2018) propose the task of cross-lingual lexical sememe prediction, aiming to extend HowNet to a new language by predicting sememes for words in that language. Qi et al (2020a) present a more efficient way to extend HowNet to other languages, i.e., building a multilingual SKB based on Babel-Net (Navigli and Ponzetto, 2012). BabelNet is composed of multilingual synsets that contain synonyms in many languages.…”

“…An obvious solution to this problem is to build an SKB for each language manually, but it is not realistic because it would be unimaginably time-consuming and laborintensive. 1 To address the challenge, previous studies try to extend HowNet to other languages by automatically predicting sememes for words in those languages (Qi et al, 2018(Qi et al, , 2020a. However, existing methods are not effective enough, and manual effort is necessary to ensure the correctness of their sememe prediction results.…”

A bilateral context and filtering strategy-based approach to Chinese entity synonym set expansion

“…To generate Chinese distant supervision knowledge, we apply 1 Hanlp 2 to link the entities involved in Cn-SynSets to two Chinese web corpora, namely, Baidu Encyclopedia articles 1 and SogouCA. 3 Next, we generate two real-world Chinese entity synonym set datasets: BDSynSetTra and SGSynSetTra, respectively, denoted as…”

“…Mining entity synonym set is an important task for many entity-based downstream applications, such as knowledge graph construction [1][2][3][4], taxonomy learning [5][6][7][8], and question answering [9][10][11]. An entity synonym set usually contains several different strings representing an identical entity [12][13][14].…”