摘要
方面术语提取是基于方面的情感分析中的一个关键的任务,其目的是从在线用户评论中提取关键的方面术语。本文通过在现实数据上使用了一个解决方面术语提取任务的新框架。该框架通过挖掘2个有用的线索,即意见摘要和方面预测的历史。意见摘要是从整个输入语句中提取出来的关键词,以每个当前标记为条件进行方面预测,因此确定的摘要可以帮助对该标记进行方面预测。另一条线索是方面预测的历史,是从以前的方面预测中提取出来的关键词,以便利用坐标结构和标注模式约束来更好地做出方面预测。用此模型分析餐馆的用户评论,最终的实验结果则展示了良好的提取结果。
Aspect term extraction is a key task in aspect-based sentiment analysis,whose purpose is to extract key aspect terms from online user reviews.This paper uses a new framework to address the aspect term extraction task by using it on real data.The framework digs through two useful threads,namely the opinion summary and the aspect forecast history.The opinion summary is a keyword extracted from the entire input statement,and the aspect prediction is based on each current tag,so the determined summary can help the aspect prediction of the tag.Another clue is the history of aspect prediction,which is the extraction of key words from previous aspect prediction in order to make better aspect prediction with coordinate structure and annotation pattern constraints.This model is used to analyze the restaurant user reviews,and the final experimental results show that the research achieves a good extraction result.
作者
雷叶
LEI Ye(School of Computer Science,Xi'an Shiyou University,Xi'an 710065,China)
出处
《智能计算机与应用》
2019年第3期259-262,共4页
Intelligent Computer and Applications
关键词
方面提取
长短期记忆网络
注意力机制
展望
aspect term extraction
Long Short-Term Memory Networks
attention mechanism
expectation
