AFET: Automatic Fine-Grained Entity Typing by Hierarchical Partial-Label Embedding

Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, pages 1369–1378, Austin, Texas, November 1-5, 2016. c⃝2016 Association for Computational Linguistics AFET: Automatic Fine-Grained Entity Typing by Hierarchical Partial-Label Embedding Xiang Ren†∗ Wenqi He†∗ Meng Qu† Lifu Huang♯ Heng Ji♯ Jiawei Han† † University of Illinois at Urbana-Champaign, Urbana, IL, USA ♯ Computer Science Department, Rensselaer Polytechnic Institute, USA †{xren7, wenqihe3, mengqu2, hanj}@illinois.edu ♯{huangl7, jih}@rpi.edu Abstract Distant supervision has been widely used in current systems of fine-grained entity typ- ing to automatically assign categories (en- tity types) to entity mentions. However, the types so obtained from knowledge bases are often incorrect for the entity mention’s local context. This paper proposes a novel em- bedding method to separately model “clean” and “noisy” mentions, and incorporates the given type hierarchy to induce loss functions. We formulate a joint optimization problem to learn embeddings for mentions and type- paths, and develop an iterative algorithm to solve the problem. Experiments on three pub- lic datasets demonstrate the effectiveness and robustness of the proposed method, with an average 15% improvement in accuracy over the next best compared method1. 1 Introduction Assigning types (e.g., person, organization) to mentions of entities in context is an important task in natural language processing (NLP). The ex- tracted entity type information can serve as primi- tives for relation extraction (Mintz et al., 2009) and event extraction (Ji and Grishman, 2008), and as- sists a wide range of downstream applications in- cluding knowledge base (KB) completion (Dong et al., 2014), question answering (Lin et al., 2012) and entity recommendation (Yu et al., 2014). While ∗Equal contribution. 1Codes and datasets used in this paper can be down- loaded at https://github.com/shanzhenren/AFET. Mention: “Schwarzenegger”; Context: S3; Candidate Type Set: {person, politician, artist, actor, author, businessman, althete} ID Sentence S1 S2 S3 ... Governor Arnold Schwarzenegger gives a speech at Mission Serve's service project on Veterans Day 2010. The fourth movie in the Predator series entitled 'The Predator' may see the return of action-movie star Arnold Schwarzenegger to the franchise. Schwarzenegger’s first property investment was a block of six units, for which he scraped together $US27,000. ... Entity: Arnold Schwarzenegger Knowledge Bases Noisy Training Examples Candidate Type Set (Sub-tree) root product person location organiz ation ... ... politician artist business man ... ... ... author actor singer ... Target Type Hierarchy Mention: “Arnold Schwarzenegger”; Context: S1; Candidate Type Set: {person, politician, artist, actor, author, businessman, althete} ... Mention: “Arnold Schwarzenegger”; Context: S2; Candidate Type Set: {person, politician, artist, actor, author, businessman, althete} S1 Distant Supervision althete S2 S3 Figure 1: Current systems may detect Arnold Schwarzenegger in sentences S1-S3 and assign the same types to all (listed within braces), when only some types are correct for context (blue labels within braces). traditional named entity recognition systems (Rati- nov and Roth, 2009; Nadeau and Sekine, 2007) fo- cus on a small set of coarse types (typically fewer than 10), recent studies (Ling and Weld, 2012; Yosef et al., 2012) work on a much …

Content truncated to 500 words