融合位置信息的观点三元组情感分析模型

方面级情感分析主要有两大类任务：a)抽取任务，旨在抽取出语句中的方面词及观点词；b)分类任务，旨在分析情感极性。在这两种复合任务的基础上，针对目前方面词与观点词耦合性较差，导致分类任务出错这一问题，提出了融合位置信息的观点三元组情感分析模型OTPM。该模型利用双向长短时记忆网络获得文本表示，接着利用自注意力机制来增强方面词与情感词之间的关联性，之后在多任务框架中进行观点三元组的抽取，同时将抽取出的表示与位置信息进行加权融合，最后利用biaffine评分器分析加权后的方面词与观点词之间的情感依赖关系，并利用stop-on-non-I算法对三元组进行解码输出三元组。在Lap14、Rest14、Rest15、Rest16四个数据集上进行大量实验，结果表明所提模型优于一系列基线模型。

Sentiment analysis model of opinion triplet based on position information

There are two kinds of tasks in aspect-based sentiment analysis: one is extraction task, which aims to extract aspect words and opinion words from sentences. The other is the classification task, which aims to analyze the sentimental polarity. On the basis of these two complex tasks, aiming at the problem that the coupling between aspect words and opinion words is poor, which leads to the error of classification task, this paper proposed an opinion triplet sentiment analysis model(OTPM) integrating position information. The model used bi-directional long short-term memory network to obtain the text representation, then used the self attention mechanism to enhance the correlation between aspect words and opinion words, and then extracted the opinion triplet in the multi task framework. At the same time, it conducted weighted fusion of extracted representation and position information. Finally, the experiment used biaffine scorer to analyze the sentimental dependence between the weighted aspect words and opinion words, and then used stop-on-non-I algorithm to decode the triplet and output the triplet. This paper conducted a lot of experiments on Lap14, Rest14, Rest15, Rest16 dataset. The results show that the proposed model is superior to a series of baseline models.