A Multi-Task Incremental Learning Framework with Category Name Embedding for Aspect-Category Sentiment Analysis

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing, pages 6955–6965, November 16–20, 2020. c⃝2020 Association for Computational Linguistics 6955 A Multi-Task Incremental Learning Framework with Category Name Embedding for Aspect-Category Sentiment Analysis Zehui Dai*, Cheng Peng, Huajie Chen, Yadong Ding NLP Group, Gridsum {daizehui,pengcheng01,chenhuajie,dingyadong}@gridsum.com Abstract (T)ACSA tasks, including aspect-category sen- timent analysis (ACSA) and targeted aspect- category sentiment analysis (TACSA), aims at identifying sentiment polarity on predefined categories. Incremental learning on new cat- egories is necessary for (T)ACSA real applica- tions. Though current multi-task learning mod- els achieve good performance in (T)ACSA tasks, they suffer from catastrophic forget- ting problems in (T)ACSA incremental learn- ing tasks. In this paper, to make multi-task learning feasible for incremental learning, we proposed Category Name Embedding network (CNE-net). We set both encoder and decoder shared among all categories to weaken the catastrophic forgetting problem. Besides the origin input sentence, we applied another in- put feature, i.e., category name, for task dis- crimination. Our model achieved state-of-the- art on two (T)ACSA benchmark datasets. Fur- thermore, we proposed a dataset for (T)ACSA incremental learning and achieved the best per- formance compared with other strong base- lines. 1 Introduction Sentiment analysis has become an increasingly popular natural language processing (NLP) task in academia and industry. It provides real-time feedback on consumer experience and their needs, which helps producers to offer better services. To deal with the presence of multiple categories in one document, (T)ACSA tasks, including aspect- category sentiment analysis (ACSA) and targeted aspect-category sentiment analysis (TACSA), were introduced. The main purpose for ACSA task is to identify sentiment polarity (i.e. positive, neutral, negative and none) of an input sentence upon specific pre- defined categories (Mohammad et al., 2018; Wu et al., 2018). For example, as shown in Table 1, giving an input sentence “Food is always fresh and hot-ready to eat, but it is too expensive.” and pre- defined categories {food, service, price, ambience and anecdotes/miscellaneous}, the sentiment of category food is positive, the polarity regarding to category price is negative, while is none for oth- ers. In this task, the models should capture both explicit expressions and implicit expressions. For example, the phrase “too expensive” indicates the negative polarity in the price category, without a direct indication of “price”. In order to deal with ACSA with both multi- ple categories and multiple targets, TACSA task was introduced (Saeidi et al., 2016) to analyze sentiment polarity on a set of predefined target- category pairs. An example is shown in Table 1, given targets “restaurant-1” and “restaurant-2”, in the case “I like restaurant-1 because it’s cheap, but restaurant-2 is too expansive”, the category price for target “restaurant-1” is positive, but is negative for target “restaurant-2”, while is none for other target-category pairs. A mathematical definition for (T)ACSA is given as follows: giving a sentence s as input, a predefined set of targets T and a pre- defined set of aspect categories A, a model predicts the sentiment polarity y for each target-category pair …

Content truncated to 500 words