TGB at SemEval-2016 Task 5: Multi-Lingual Constraint System for Aspect Based Sentiment Analysis

Proceedings of SemEval-2016, pages 337–341, San Diego, California, June 16-17, 2016. c⃝2016 Association for Computational Linguistics TGB at SemEval-2016 Task 5: Multi-Lingual Constraint System for As- pect Based Sentiment Analysis Fatih Samet Çetin Istanbul Technical University cetinfatih@itu.edu.tr Ezgi Yıldırım Istanbul Technical University yildirimez@itu.edu.tr Can Özbey Turkcell Global Bilgi canozbeycan@gmail.com Gülşen Eryiğit Istanbul Technical University gulsen.cebiroglu@itu.edu.tr Abstract This paper gives the description of the TGB system submitted to the Aspect Based Senti- ment Analysis Task of SemEval-2016 (Task 5). The system is built on linear binary classifiers for aspect category classification (Slot 1), on lexicon-based detection for opinion target ex- pressions extraction (Slot 2), and on linear multi-class classifiers for sentiment polarity detection (Slot 3). We conducted several differ- ent approaches for feature selection to improve classification performance on both Slot 1 and Slot 3. Our proposed methods are easily adapt- able to all languages and domains since they are built as constrained systems which do not use any additional resources other than the pro- vided datasets and which uses standard prepro- cessing methods. 1 Introduction Since Web 2.0 and social media platforms have be- come popular in recent times, the amount of acces- sible text data has shown rapid increase. The man- ual analysis of this huge amount of data is almost impossible to accomplish in a reasonable time, thus automatic sentiment analysis and opinion mining have turned into a significant requirement for com- panies. The most of earlier studies conducted in this area were generally focused on document level (Yıldırım, et al., 2015; Pang, et al., 2002; Esuli & Sebastiani, 2006) until the recognition of that dif- ferent opinions can exist in the same sentence or paragraphs. Another disadvantage of general senti- ment analysis approaches is the disability to match the sentiment polarities to the target entities. Therefore, this type of analysis becomes insuffi- cient for deep understanding of opinions about products and features. The most commonly refer- enced study (Liu, 2012) on Aspect-Based Senti- ment Analysis (ABSA) discusses the problem as extracting the tuples including multiple opinions.. The need for a detailed sentiment analysis with re- spect to specific target entities has given birth to ABSA. In International Workshop on Semantic Evaluation (SemEval), a shared task called Aspect- based Sentiment Analysis has been actualized since 2014 (Pontiki, et al., 2014; Pontiki, et al., 2015; Pontiki, et al., 2016). In this year’s task, the data is annotated at both sentential and textual lev- els with reference to predefined domain-dependent aspect categories. For more information and details about the aspect categories consult (Pontiki et al., 2016). The task description (SemEval, 2016) pro- vides regulations as to how these categories should be determined. This paper presents our system prepared for ABSA 2016. It covers 3 different languages, namely, English, Spanish and Dutch. The system has multi-lingual capabilities giving nearly con- 337 sistent performances for each language. Our pro- posed methods are easily adaptable to all languages due to their multi-lingual nature by switching lan- guage codes for stemming phase. Considering the …

Content truncated to 500 words