Question-Answering-system-using-Cosine-similarity

In this project we will find the best possible answer to the multiple choice question using tfidf embeddings and cosine similarity.

Part 1 - Model:

• We will build the term document matrix using data.txt.
• Treat each sentence as single document
• Term document matrix will contain tf-idf values.

Part 2 - Query:

• We will be using test.jsonl file to obtain the questions.
• Each line contains 1 question in json format along with 4 options.
• To form a query, combine each question with it’s option, ie, (Q​i​ + O​i​).

Evaluation :

• For each query, we find the cosine similarity between the query and each of the documents and use the similarity score of the most matching document as the score of your system for the query.
• We repeat this for all the options and our answer for a particular question should be the (Q​i + O​i​) combination with the maximum score.