VLH team at ALQAC 2022: Retrieving legal document and extracting answer with BERT-based model

Hai-Long Nguyen, Thai-Binh Nguyen, Tan-Minh Nguyen, Ha-Thanh Nguyen, Hai-Yen Thi Vuong

Abstract: Legal AI is a growing subfield of AI that deals with the application of AI techniques to legal problems. In legal AI, there are many different tasks that can be performed, such as legal retrieval, legal question answering, and legal document generation. In this paper, we describe our approach to the legal retrieval and legal question answering tasks of the Automated Legal Question Answering Competition (ALQAC) 2022. For legal retrieval, we use a combination of lexical matching with deep learning models. For legal question answering, we treat the inputs in multiple aspects and come up with different strategies to answer the questions. Our system achieves impressive results in both tasks.