Building a Question and Answer System for News Domain
This work addresses the problem of automated question answering in the news domain for users seeking quick information, but it is incremental as it applies existing methods like BERT to this specific context.
This project tackled building a question-answering system for news articles by developing span-based models with attention mechanisms, achieving an F1 score of 33.095 with a Bi-LSTM model and 57.513 with a BERT-based model on the SQuAD 2.0 dataset.
This project attempts to build a Question- Answering system in the News Domain, where Passages will be News articles, and anyone can ask a Question against it. We have built a span-based model using an Attention mechanism, where the model predicts the answer to a question as to the position of the start and end tokens in a paragraph. For training our model, we have used the Stanford Question and Answer (SQuAD 2.0) dataset[1]. To do well on SQuAD 2.0, systems must not only answer questions when possible but also determine when no answer is supported by the paragraph and abstain from answering. Our model architecture comprises three layers- Embedding Layer, RNN Layer, and the Attention Layer. For the Embedding layer, we used GloVe and the Universal Sentence Encoder. For the RNN Layer, we built variations of the RNN Layer including bi-LSTM and Stacked LSTM and we built an Attention Layer using a Context to Question Attention and also improvised on the innovative Bidirectional Attention Layer. Our best performing model which uses GloVe Embedding combined with Bi-LSTM and Context to Question Attention achieved an F1 Score and EM of 33.095 and 33.094 respectively. We also leveraged transfer learning and built a Transformer based model using BERT. The BERT-based model achieved an F1 Score and EM of 57.513 and 49.769 respectively. We concluded that the BERT model is superior in all aspects of answering various types of questions.