Learning entity-centric document representations using an entity facet topic model
作者：
Highlights：
• We propose the task of entity-centric document representation learning.
• We propose a novel Entity Facet Topic Model (EFTM) to learn entity-centric document representations.
• We confirm our hypothesis regarding the existence of multiple facets of an entity by analysing the learned entity facets using qualitative and quantitative analysis, and identify a effective number of facets per entity.
• We demonstrate the effectiveness of EFTM in downstream applications using a multilabel classification task.
摘要
•We propose the task of entity-centric document representation learning.•We propose a novel Entity Facet Topic Model (EFTM) to learn entity-centric document representations.•We confirm our hypothesis regarding the existence of multiple facets of an entity by analysing the learned entity facets using qualitative and quantitative analysis, and identify a effective number of facets per entity.•We demonstrate the effectiveness of EFTM in downstream applications using a multilabel classification task.
论文关键词：Document representation,Topic models,Entity aspects,Text classification
论文评审过程：Received 30 September 2018, Revised 23 January 2020, Accepted 26 January 2020, Available online 12 February 2020, Version of Record 12 February 2020.
论文官网地址：https://doi.org/10.1016/j.ipm.2020.102216
