Dead or Murdered? Predicting Responsibility Perception in Femicide News Reports

Proceedings of the 2nd Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics and the 12th International Joint Conference on Natural Language Processing (Volume 1: Long Papers), pages 82–92 November 20–23, 2022. ©2022 Association for Computational Linguistics 82 Graph-augmented Learning to Rank for Querying Large-scale Knowledge Graph Hanning Gao1∗, Lingfei Wu2∗, Po Hu3†, Zhihua Wei1†, Fangli Xu4 and Bo Long5 1Tongji University, 2Pinterest, 3Central China Normal University 4Squirrel AI Learning, 5JD.COM gaohn@tongji.edu.cn, lwu@email.wm.edu phu@mail.ccnu.edu.cn, zhihua_wei@tongji.edu.cn lili@yixue.us, bo.long@jd.com Abstract Knowledge graph question answering (KGQA) based on information retrieval aims to answer a question by retrieving answer from a large- scale knowledge graph. Most existing methods first roughly retrieve the knowledge subgraphs (KSG) that may contain candidate answer, and then search for the exact answer in the KSG. However, the KSG may contain thousands of candidate nodes since the knowledge graph in- volved in querying is often of large scale, thus decreasing the performance of answer selection. To tackle this problem, we first propose to par- tition the retrieved KSG to several smaller sub- KSGs via a new subgraph partition algorithm and then present a graph-augmented learning to rank model to select the top-ranked sub-KSGs from them. Our proposed model combines a novel subgraph matching networks to cap- ture global interactions in both question and subgraphs, and an Enhanced Bilateral Multi- Perspective Matching model is proposed to cap- ture local interactions. Finally, we apply an answer selection model on the full KSG and the top-ranked sub-KSGs respectively to vali- date the effectiveness of our proposed graph- augmented learning to rank method. The exper- imental results on multiple benchmark datasets have demonstrated the effectiveness of our ap- proach. 1 Introduction With the rise of large-scale knowledge graphs (KG) such as DBpedia (Auer et al., 2007) and Freebase (Bollacker et al., 2008), question answering over knowledge graph has attracted massive attention re- cently, which aims to leverage the factual informa- tion in a KG to answer natural language question. Depending on the complexity of question, KGQA can be divided into two forms: simple and com- plex. Simple KGQA often requires only one hop of factual knowledge, while complex KGQA requires *These authors contributed equally to this work. †Corresponding authors. reasoning over a multi-hop knowledge subgraph (KSG) and selecting the correct answer among sev- eral candidate answers. In this paper, we focus on the latter, i.e., complex KGQA, which is more challenging. Currently, most KGQA approaches resort to se- mantic parsing (Berant et al., 2013; Yih et al., 2015; Dong and Lapata, 2018) or retrieve-then-extract methods (Yao and Van Durme, 2014; Bordes et al., 2014). Semantic parsing methods usually translate a natural language question to a KG query and then use it to query the KG directly. However, semantic parsing methods often rely on complex and spe- cialised hand-crafted rules or schemes. In contrast, retrieve-then-extract methods are easier to under- stand and more interpretable. They first retrieve the KG coarsely to obtain a KSG containing answer candidates. Then, the target answer is extracted …

Content truncated to 500 words