Improving Entity Linking through Semantic Reinforced Entity Embeddings

Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 6843–6848 July 5 - 10, 2020. c⃝2020 Association for Computational Linguistics 6843 Improving Entity Linking through Semantic Reinforced Entity Embeddings Feng Hou♯†, Ruili Wang♯†∗, Jun He‡, Yi Zhou♮† ♯School of Computer and Information Engineering, Zhejiang Gongshang University, Hangzhou, China †School of Natural and Computational Sciences, Massey University, New Zealand ‡School of Information Communication, National University of Defense Technology, China ♮Shanghai Research Center for Brain Science and Brain-Inspired Intelligence, Zhangjiang Lab, China {f.hou, ruili.wang}@massey.ac.nz, prof.ruili.wang@gmail.com hejun nudt@nudt.edu.cn, yzhou@bsbii.cn Abstract Entity embeddings, which represent different aspects of each entity with a single vector like word embeddings, are a key component of neural entity linking models. Existing en- tity embeddings are learned from canonical Wikipedia articles and local contexts surround- ing target entities. Such entity embeddings are effective, but too distinctive for linking models to learn contextual commonality. We propose a simple yet effective method, FGS2EE, to in- ject fine-grained semantic information into en- tity embeddings to reduce the distinctiveness and facilitate the learning of contextual com- monality. FGS2EE first uses the embeddings of semantic type words to generate semantic embeddings, and then combines them with ex- isting entity embeddings through linear aggre- gation. Extensive experiments show the effec- tiveness of such embeddings. Based on our entity embeddings, we achieved new state-of- the-art performance on entity linking. 1 Introduction Entity Linking (EL) or Named Entity Disambigua- tion (NED) is to automatically resolve the ambi- guity of entity mentions in natural language by linking them to concrete entities in a Knowledge Base (KB). For example, in Figure 1, mentions “Congress” and “Mr. Mueller” are linked to the corresponding Wikipedia entries, respectively. Neural entity linking models use local and global scores to rank and select a set of entities for men- tions in a document. Entity embeddings are critical for the local and global score functions. But the current entity embeddings (Ganea and Hofmann, 2017) encoded too many details of entities, thus are too distinctive for linking models to learn con- textual commonality. ∗Corresponding author extracting�fine-grained�semantic�types exonerated the president of obstruction of justice. [legislature,�government,�u.s.] [american,�lawyer,�government,�official] [european,�assembly] [german, canadian, poker, player] Congress�of�the�Council�of�Europe United�States�Congress Robert�Mueller Greg Mueller wikipedia�articles Appearing before Congress, Mr Mueller said he had not Figure 1: Entity linking with embedded fine-grained semantic types We hypothesize that fine-grained semantic types of entities can let the linking models learn contex- tual commonality about semantic relatedness. For example, rugby related documents would have en- tities of rugby player and rugby team. If a linking model learns the contextual commonality of rugby related entities, it can correctly select entities of similar types using the similar contextual informa- tion. In this paper, we propose a method FGS2EE to inject fine-grained semantic information into entity embeddings to reduce the distinctiveness and facilitate the learning of contextual commonality. FGS2EE uses the word embeddings of semantic words that represent the hallmarks of entities (e.g., writer, carmaker) to generate semantic embeddings. We find that the training converges faster when using …

Content truncated to 500 words