IHS-RD-Belarus at SemEval-2016 Task 5: Detecting Sentiment Polarity Using the Heatmap of Sentence

Proceedings of SemEval-2016, pages 296–300, San Diego, California, June 16-17, 2016. c⃝2016 Association for Computational Linguistics IHS-RD-Belarus at SemEval-2016 Task 5: Detecting Sentiment Polarity Using the Heatmap of Sentence Maryna Chernyshevich IHS Inc. / IHS Global Belarus 131 Starovilenskaya St 220123, Minsk, Belarus {Marina.Chernyshevich}@ihs.com Abstract This paper describes the system submitted by IHS-RD-Belarus team for the sentiment detec- tion polarity subtask on Aspect Based Senti- ment Analysis task at the SemEval 2016 workshop on semantic evaluation. We devel- oped a system based on artificial neural net- work to detect the sentiment polarity of opin- ions. Evaluation on the test data set showed that our system achieved the F-score of 0.83 for restaurants domain (rank 4 out of 28 sub- missions) and F-score of 0.78 for laptops do- main (rank 4 out of 21 submissions). 1 Introduction Social media texts found in user review services have a great data-mining potential, as they offer real-time data that can be useful to monitor public opinion on brands, products, events, etc. Most of recent approaches to sentiment analysis task are based on bag-of-words features, syntactic dependency features, out-of-domain and domain- specific sentiment lexicons, to train an supervised model that predicts the polarity of each given term or aspect category. This approach is very popular but it relies on heavy pre-processing of the data which involves careful choosing of the right fea- tures, empirical thresholds and intuitive analysis of the training set (Brun et al., 2014, Saias, 2015). In this paper, we present an approach to the opinion polarity detection task based on an artifi- cial neural network and sentiment orientation score of words. 2 Task description The SemEval-2016 shared task on Aspect based Sentiment Analysis focuses on identifying the Opinion target expressions (OTE), the Aspect cat- egories and the sentiment expressed towards each OTE or Aspect category. The main focus of this paper is polarity sub- tasks, such as OTE polarity in restaurant subject domain and Aspect category polarity in laptop sub- ject domain. In the OTE polarity subtask, the input consists of a review sentence and the set of terms or aspect categories. The expected output is a polarity label (positive, negative or neutral) for each of the asso- ciated terms or aspect categories. For example, the system should determine the polarity of fajitas and salads in the following sen- tence: I hated their fajitas, but their salads were great. As for the Aspect category polarity, the task is more complicated. In the following sentence, the system has to determine the polarity of display quality (DISPLAY#QUALITY) and display usa- bility (DISPLAY#USABILITY): The display has a great resolution but has difficulty always seeing the small print. The task organizers provided a dataset of cus- tomer reviews with manually annotated opinion targets: 2500 sentences for laptop domain and 2000 sentences for restaurant domain. Evaluation was to be carried out according to Precision, Recall and F1 metrics. 296 3 System description The central idea behind our system is the visualiza- tion of sentiment orientation in a …

Content truncated to 500 words