Dataset Creation and Evaluation of Aspect Based Sentiment Analysis in Telugu, a Low Resource Language

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 5017–5024 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 5017 Dataset Creation and Evaluation of Aspect Based Sentiment Analysis in Telugu, a Low Resource Language Regatte Yashwanth Reddy, Gangula Rama Rohit Reddy, Radhika Mamidi LTRC IIIT Hyderabad Hyderabad Telangana India yashwanthreddy.r@research.iiit.ac.in, ramarohitreddy.g@research.iiit.ac.in, radhika.mamidi@iiit.ac.in Abstract In recent years, sentiment analysis has gained popularity as it is essential to moderate and analyse the information across the internet. It has various applications like opinion mining, social media monitoring, and market research. Aspect Based Sentiment Analysis (ABSA) is an area of sentiment analysis which deals with sentiment at a finer level. ABSA classifies sentiment with respect to each aspect to gain greater insights into the sentiment expressed. Significant contributions have been made in ABSA, but this progress is limited only to a few languages with adequate resources. Telugu lags behind in this area of research despite being one of the most spoken languages in India and an enormous amount of data being created each day. In this paper, we create a reliable resource for aspect based sentiment analysis in Telugu. The data is annotated for three tasks namely Aspect Term Extraction, Aspect Polarity Classification and Aspect Categorisa- tion. Further, we develop baselines for the tasks using deep learning methods demonstrating the reliability and usefulness of the resource. Keywords: Aspect based Sentiment Analysis, Indian Languages, Deep Learning, Machine Learning 1. Introduction The emergence and rapid growth of the internet made the availability of information easy. The availability of this enormous amount of information posed several challenges which led to the development of new research areas. Sen- timent Analysis is one of them. Internet powers users with the freedom of expressing their views through blogs and social media platforms. Processing and analysing this in- formation helps in attaining an adequate understanding of customers’ opinions. Sentiment Analysis (SA) enables ef- fective judgment of a product/service by assigning polari- ties to the reviews given by its users. Sentiment analysis can be performed at different levels. As- pect Based Sentiment analysis (ABSA) is a finer level sen- timent analysis that assigns polarity to each targeted aspect instead of the entire review to increase the level of under- standing. An Aspect can be a property or component of the product/service. ABSA involves the tasks of (a) Identi- fying the Aspect terms in the sentences of the user review and (b) Assigning polarity to the identified Aspect terms. There can be more than one Aspect terms in a sentence. Many approaches are developed to assess ABSA and its challenges in recent works. But most of these works are limited only to a few languages, where a majority of them are in English. Very few of them focus on Indian lan- guages. Despite Telugu being the fourth most spoken lan- guage in India and sixteenth across the world, the area of aspect based sentiment analysis in Telugu remains unex- plored. There are several challenges …

Content truncated to 500 words