Predicting Subjective Features from Questions on QA Websites using BERT
share
Modern Question-Answering websites, such as StackOverflow and Quora, have specific user rules to maintain their content quality. These systems rely on user reports for accessing new contents, which has serious problems including the slow handling of violations, the loss of normal and experienced users' time, the low quality of some reports, and discouraging feedback to new users. Therefore, with the overall goal of providing solutions for automating moderation actions in Q A websites, we aim to provide a model to predict 20 quality or subjective aspects of questions in QA websites. To this end, we used data gathered by the CrowdSource team at Google Research in 2019 and fine-tuned pre-trained BERT model on our problem. Model achieves 95.4 epochs of training and did not improve substantially in the next ones. Results confirm that by simple fine-tuning, we can achieve accurate models, in little time, and on less amount of data.
READ FULL TEXT