• + 0 comments

    I achieved a score of 34 using TF-IDF with cosine_similarity from scikit-learn. I fit the TF-IDF vectorizer on the context text, questions, and answers, and then transformed each sentence in the context individually, along with the questions and answers. After that, I computed the cosine similarity between each sentence vector and the question vectors, as well as between each sentence vector and the answer vectors. Finally, I took the product of both similarity scores to evaluate relevance.