Biomedical entity linking based on less labeled data  

作　　者：Yu HU Derong SHEN Tiezheng NIE Yue KOU Ge YU 

机构地区：[1]School of Computer Science and Engineering,Northeastern University,Shenyang 110819,China

出　　处：《Frontiers of Computer Science》2022年第3期219-221,共3页中国计算机科学前沿（英文版）

基　　金：the National Natural Science Foundation of China(62072084,62072086,62172082);the National Defense Basic Scien tific Research Program of China(JCKY2018205C012);the Fundamental Research Funds for the Central Universities(N2116008).

摘　　要：1 Introduction Biomedical entity linking aims to align the natural language entity to the knowledge base concept referring to the same real-world object.Recent solutions in the biomedical field focus on the embedding-based method to jointly model the texts and the knowledge graph into a multi-dimensional entity space.However,the task of a biomedical entity linking is still facing tough challenges:1)The ambiguity of natural language descriptions,including polysemy and abbreviation.

关 键 词：ABBREVIATION TOUGH AMBIGUITY 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]