An Aspect-Level Sentiment Analysis Dataset for Therapies on Twitter

This dataset is an aspect-level sentiment analysis dataset for therapies, created by leveraging user-generated text from Twitter. The dataset contains a total of 5364 tweets related to 32 chronic pain therapies. These tweets are further categorized into 998 (18.6%) positive, 619 (11.5%) negative, and 3747 (69.9%) neutral sentiments. The inter-annotation agreement for the dataset was evaluated using Cohen's Kappa score, achieving an 0.82 score.

At the time of submission, this dataset is used for SMM4H 2023 shared tasks. The labels of the test set will be added after the shared task. More details about the shared task can be found at https://healthlanguageprocessing.org/smm4h-2023/. In addition, we only publicly provide the tweet IDs instead of the text content which is required by the Twitter privacy policy. Please download the text content using the tweet IDs via the Twitter API or reach out to yuting.guo@emory.edu.