A Benchmark Dataset for Sentiment Analysis of Users' Reviews on COVID-19 Contact Tracing Applications

This dataset is intended to support sentiment analysis of users' reviews on COVID-19 contact tracing mobile applications. The dataset is composed of a total of 34,53 manually annotated users' reviews from 46 different applications used in different parts of the world. These reviews are categorized into 4 categories including positive, negative, neutral, and technical issues. The positive class is composed of 15,587 reviews, 8,178 reviews are in the negative class while the neutral and technical issues classes are composed of 1,271 and 9,496, respectively. The data is intended to cover three different tasks. The details of each task are provided in the paper. Moreover, in order to facilitate the potential users of the dataset, we are providing the task-wise distribution of reviews for each task in a separate folder. Each folder provides separate test and train files each with three columns representing sequence number, text, and label. Some details of the labels in each task are provided below Task 1: Positive reviews = 0, negative reviews = 1, Technical Issues = 2 Task 2 Positive reviews = 0, Negative reviews = 1 Task 3 Positive reviews = 0, negative reviews =1, neutral = 2 Please cite the following paper, if you use the dataset in your work