[Eniafe Festus Ayetiran] Exploting Semantic Knowledge for Aspect Sentiment Classification: A Deep Learning Approach 3. 12. 2020
Abstract
Aspect Based Sentiment Analysis (ABSA) is the mining of opinions from a text about specific entities and their aspects. ABSA can provide both consumers and businesses insights to help in decision making. Currently, the state-of-the-art techniques are supervised. However, training data annotation is cumbersome and takes a lot of time to develop. Furthermore, there are several aspects and a huge amount of opinionated texts available which makes it more difficult to develop any training data
that will scale on real-life data. Besides, for any supervised system to perform very well, training data must be related to the test data. Aspect-based sentiment Analysis can benefit from heuristics derived from a combination of semantic knowledge and syntax knowledge. Recently, external document knowledge has been used to enhanced supervised sentiment classification using the deep learning technique. The aim of this work is to incorporate semantic and document knowledge into an LSTM-based approach.
References for Readings
- Soujanya Poria, Erik Cambria, Lun-Wei Ku, Chen Gui, and Alexander Gelbukh. (2014). A Rule-Based Approach to Aspect Extraction from Product Reviews. In: Proceedings of the Second Workshop on Natural Language Processing for Social Media (SocialNLP), pp. 28–37.
- Yequan Wang and Minlie Huang and Li Zhao and Xiaoyan Zhu. (2016) Attention-based LSTM for Aspect-level Sentiment Classification. In: Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing (EMNLP 2016), pp. 606–615.
- Kateřina Veselovská, Aleš Tamchyna. (2014). ÚFAL: Using Hand-crafted Rules in Aspect Based Sentiment Analysis on Parsed Data. In: Proceedings of the 8th International Workshop on Semantic Evaluation (SemEval 2014), pp. 694–698.
- Pontiki et al. (2016). SemEval-2016 Task 5: Aspect Based Sentiment Analysis. In: Proceedings of SemEval-2016. pp. 19–30.
