Simhasth 2016 Feedback Dataset: A Domain Specific Sentiment Analysis Dataset and its Baseline Evaluation

Sentiment analysis is a very important task in Natural Language Processing. Sentiment analysis is intended to find the emotion or feedback given in some review or text. Sentiment analysis is intensively explored in resource-rich languages like English, but for languages like Hindi there are only few resources for sentiment analysis studies. Here, a novel dataset for sentiment analysis in Hindi language is introduced, created from feedback on Simhasth 2016. Simhasth is a very important and huge spiritual and religious gathering of Hindu religious fellows held in Ujjain city of Madhya Pradesh in India. We collected feedback on the arrangements and experiences of visitors of Simhasth 2016. Feedbacks were originally collected as text on feedback paper sheets, then feedback texts were digitized as a feedback dataset. From this feedback dataset, after annotation and class balancing, we finally got the proposed dataset. We also performed some baseline performance evaluations on the dataset and compared various machine learning techniques for sentiment analysis. We used various machine learning techniques for evaluation. This feedback dataset has almost all feedback in Hindi. The dataset is now available as a resource for sentiment analysis in the Hindi language<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">*</sup>.