An Empirical Study on Multiple Information Sources for Zero-Shot Fine-Grained Entity Typing

Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 2668–2678 November 7–11, 2021. c⃝2021 Association for Computational Linguistics 2668 An Empirical Study on Multiple Information Sources for Zero-Shot Fine-Grained Entity Typing Yi Chen1, Haiyun Jiang1, Lemao Liu1, Shuming Shi1, Chuang Fan Min Yang2, Ruifeng Xu3∗ 1Tencent AI Lab, Shenzhen, China 2Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences 3Peng Cheng Laboratory, Shenzhen, China yichennlp@gmail.com, haiyunjiang@tencent.com lemaoliu@gmail.com, shumingshi@tencent.com fanchuanghit@gmail.com, min.yang@siat.ac.cn xuruifeng.hitsz@gmail.com Abstract Auxiliary information from multiple sources has been demonstrated to be effective in zero- shot fine-grained entity typing (ZFET). How- ever, there lacks a comprehensive understand- ing about how to make better use of the ex- isting information sources and how they af- fect the performance of ZFET. In this paper, we empirically study three kinds of auxiliary information: context consistency, type hierar- chy and background knowledge (e.g., proto- types and descriptions) of types, and propose a multi-source fusion model (MSF) targeting these sources. The performance obtains up to 11.42% and 22.84% absolute gains over state- of-the-art baselines on BBN and Wiki respec- tively with regard to macro F1 scores. More importantly, we further discuss the characteris- tics, merits and demerits of each information source and provide an intuitive understanding of the complementarity among them. 1 Introduction Fine-grained entity typing (FET) aims to detect the types of an entity mention given its context (Abhishek et al., 2017; Xu and Barbosa, 2018; Jin et al., 2019). The results of FET benefit lots of downstream tasks (Chen et al., 2020; Hu et al., 2019; Zhang et al., 2020a; Liu et al., 2021; Chu et al., 2020). In many scenarios, the type hierar- chy is continuously evolving, which requires newly emerged types to be accounted into FET systems. As a result, zero-shot FET (ZFET) is welcomed to handle the new types which are unseen during training stage (Ma et al., 2016; Ren et al., 2020; Zhang et al., 2020b). The major challenge of ZFET is to build the se- mantic connections between the seen types (during training) and the unseen ones (during inference). ∗Corresponding Author CA HA KA <Northwest and Midway are two of the …, /organization/corporation> 𝓛𝓛𝑪𝑪𝑪𝑪 𝓛𝓛𝑯𝑯𝑪𝑪 𝓛𝓛𝑲𝑲𝑪𝑪 Fusion Overall Loss [Mention] [Context] [Type] Source1: Context Consistency Consistent: Corporation and Midway are two of the five airlines with which Budget has agreements. Inconsistent: Drug and Midway are two of the five airlines with which Budget has agreements. Source2: Type Hierarchy Organization Government Corporation … … Source3: Background Knowledge Prototypes: western_union, quebecor, merrill, rtc, … Description: a business firm whose articles of incorporation have been approved in some state. Figure 1: Illustration of the proposed multi-source fu- sion model (MSF). Auxiliary information has been proved to be es- sential in this regard (Xian et al., 2019), with a variety of approaches focused on scattered infor- mation (Ma et al., 2016; Zhou et al., 2018; Obeidat et al., 2019; Ren et al., 2020; Zhang et al., 2020b). However, the power of auxiliary information has not been sufficiently exploited …

Content truncated to 500 words