An Extended Attention-based LSTM with Knowledge Embedding for Aspect-level Sentiment Analysis

Bo Hu ； Yun Liu ； Ming-dong Sun ； Kun Mi

《電腦學刊》 30卷4期 (2019/08) Pp. 176-184

https://doi.org/10.3966/199115992019083004017

摘要

Aspect-level sentiment analysis is a highly concerned content of sentiment analysis in recent years. For sentiment analysis, however, some of the necessary external knowledge such as commonsense is very important, and these models can't acquire this knowledge through training data. To this end, in this paper we propose an extension of the LSTM based on attention mechanism to combine explicit knowledge with tacit knowledge to improve the accuracy of the aspect-level sentiment analysis model.

關鍵字

sentiment analysis ； attention model ； commonsense knowledge