Boosting drug named entity recognition using an aggregate classifier
作者：
Highlights：
• We investigate drug NER using limited or no manually annotated data.
• We propose an algorithm for combining methods based on annotations and dictionaries.
• We improved drug NER recall using suffix patterns evolved by genetic programming.
• We improved drug NER performance by aggregating heterogenous drug NER methods.
• We conclude that drug NER can be performed competitively without manual annotations.
摘要
Highlights•We investigate drug NER using limited or no manually annotated data.•We propose an algorithm for combining methods based on annotations and dictionaries.•We improved drug NER recall using suffix patterns evolved by genetic programming.•We improved drug NER performance by aggregating heterogenous drug NER methods.•We conclude that drug NER can be performed competitively without manual annotations.
论文关键词：Named entity annotation sparsity,Gold-standard vs. silver-standard annotations,Named entity recogniser aggregation,Genetic-programming-evolved string-similarity patterns,Drug named entity recognition
论文评审过程：Available online 17 June 2015, Version of Record 19 October 2015.
论文官网地址：https://doi.org/10.1016/j.artmed.2015.05.007
