This is the code of my dissertation supervised by Prof. Hossein Sameti at CS@Sharif University of Technology and also the code of my solution submitted to SemEval 2022 workshop.
Data folder includes SemEval 2022 Task 10 data and SemEval 2014-2015 (Pontiki et al. 2014, 2015) data annotated by (Wang et al. 2016, 2017). The latter one is on Targeted Sentiment Analysis task and I obtained it from (Chen et Qian 2020) github repo.
Three notebooks are provided by this repository. Each one is explained below.
This is my proposed solution for SemEval 2022 Task 10 Subtask 1 (Structured Sentiment Analysis). It includes training, evaluation and all experiments.
Running and evaluating SemEval 2022 Task 10 baselines using its codebase in order to reproduce the results and compare my solution with them.
My implementation of a baseline for ABSA (Aspect-Based Sentiment Analysis) task. This notebook includes training and evaluation of the baseline. I used the same data that this baseline used which is available in its repo. This baseline is not among my solution baselines but I implemented it as a part of my literature study step.