Aspect-Based Sentiment Analysis as Fine-Grained Opinion Mining

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 6804–6811 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 6804 Aspect-Based Sentiment Analysis as Fine-Grained Opinion Mining Gerardo Ocampo Diaz1, Xuanming Zhang2, Vincent Ng1 1Human Language Technology Research Institute, University of Texas at Dallas 2University of Nottingham, Ningbo China godiaz@hlt.utdallas.edu, zy21855@nottingham.edu.cn, vince@hlt.utdallas.edu Abstract We show how the general fine-grained opinion mining concepts of opinion target and opinion expression are related to aspect-based sentiment analysis (ABSA) and discuss their benefits for resource creation over popular ABSA annotation schemes. Specifically, we first discuss why opinions modeled solely in terms of (entity, aspect) pairs inadequately captures the meaning of the sentiment originally expressed by authors and how opinion expressions and opinion targets can be used to avoid the loss of information. We then design a meaning-preserving annotation scheme and apply it to two popular ABSA datasets, the 2016 SemEval ABSA Restaurant and Laptop datasets. Finally, we discuss the importance of opinion expressions and opinion targets for next-generation ABSA systems. We make our datasets publicly available for download. Keywords: opinion mining, sentiment analysis, text mining 1. Introduction For almost two decades, researchers in the text mining and natural language processing (NLP) communities have worked to improve the state of aspect-based sentiment anal- ysis (ABSA); a task that, roughly speaking, involves ex- tracting sentiment/opinions from text in terms of targets they address. For example, given the sentence: The soup here is very expensive. The goal of ABSA is to extract the subject matter of the opinion, which is typically represented by an (entity,aspect) pair along with the sentiment expressed towards it (fre- quently in terms of polarity). For example, the previous sentence expresses a positive polarity towards (food, price). The entity and the aspect are both chosen from predefined sets. So, if an opinion is expressed on an entity that does not appear in the predefined set of entities, then an ABSA system is not expected to output any opinion for that entity. The important role this task can play in decision making for governments, companies, and individuals has resulted in a large body of research on the subject being devel- oped over the years, along with the organization of mul- tiple shared tasks in SemEval (Pontiki et al., 2014; Pontiki et al., 2015; Pontiki et al., 2016), TASS (Mart´ınez C´amara et al., 2018), and GermanEval (Wojatzki et al., 2017). The datasets published in the SemEval ABSA tasks (Pontiki et al., 2016), which are composed of user-generated reviews of products and services such as those found on Amazon and Yelp, have heavily influenced work in the area, as re- searchers frequently rely on them for system training and testing. However, these datasets follow a task-specific an- notation scheme, under which opinion annotations do not always capture the full meaning of the sentiment expressed in text. Consider the following pair of sentences taken from the SemEval 2016 Restaurants dataset: (1) I tend to judge a sushi restaurant by …

Content truncated to 500 words