Project Title:

Information Retrieval Approaches using Legal Text Documents

Description:

In our country, when lawyers get a new case, they compare it with the previous one. He has to do a lot of case studies to find the solution of a new case. There are also a lot of cases and it is very difficult to find a new caustic match with the previous ones.

In this project, the previous case can be found very easily which is related to the new case using information retrieval systems.

Dataset:

I used 2 datasets. One is “prior cases” that I used to train the model and the other is “current cases” that I used to find the similarity with the prior cases.

Model:

I used word2vec, Fasttext, BERT pre-trained model (bert-base-nli-mean-tokens), and DistilBERT pre-trained model(distilbert-base-nli-mean-tokens).

Result:

The final result shows the top 5 most similar prior cases of each current case.

Language:

Python

IDE:

Google Colab