FlairNLP at SemEval-2023 Task 6b: Extraction of Legal Named Entities from Legal Texts using Contextual String Embeddings

Abstract: Indian court legal texts and processes are essential towards the integrity of the judicial system and towards maintaining the social and political order of the nation. Due to the increase in number of pending court cases, there is an urgent need to develop tools to automate many of the legal processes with the knowledge of artificial intelligence. In this paper, we employ knowledge extraction techniques, specially the named entity extraction of legal entities within court case judgements. We evaluate several state of the art architectures in the realm of sequence labeling using models trained on a curated dataset of legal texts. We observe that a Bi-LSTM model trained on Flair Embeddings achieves the best results, and we also publish the BIO formatted dataset as part of this paper.