
A Joint Learning Information Extraction Method Based on an Effective Inference Structure
Author(s) -
Shiping Ma,
Xiong Chen
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1487/1/012009
Subject(s) - computer science , inference , information extraction , artificial intelligence , knowledge graph , relationship extraction , graph , joint (building) , task (project management) , benchmark (surveying) , natural language processing , machine learning , natural language , domain (mathematical analysis) , question answering , theoretical computer science , mathematics , architectural engineering , mathematical analysis , management , geodesy , engineering , economics , geography
Over the past few years, natural language processing is getting much attraction from more scholars and institutions. Knowledge graph has been regarded as a crucial role in pushing natural language understanding forward. The task of information extraction is the first step to build a large-scale knowledge graph, which means to identify information from the natural language text and extract it in the form of entity and relation triplets. Some joint learning method have been proposed in this domain recently. In this paper, we inherit the idea of joint learning, use a simple, lightweight but effective structure to solve this task and compare our method with some recent algorithms on the benchmark dataset NYT and WebNLG. Results show that our method can get an improvement in F1 score.