A Sentimental Analysis of Legal Documents Using Mask Attention BERT Networks

Legal systems can function more efficiently by processing cases faster and having a higher case clearance rate when complex legal texts are automatically analysed for logical patterns. The most crucial task in doing this is classifying sentences in legal documents automatically based on their content. This chapter suggests a deep learning model for sentiment analysis-based legal text analysis and judgment generation. The transformer model is an innovative encoder-decoder that uses self-awareness to analyse speech patterns which runs noticeably quicker and allows for parallel processing. In this work, the glove embedding and the BERT algorithm—bidirectional encoder representation for transformer model—are utilised to construct sentiment analysis for text categorization. Prior to extracting valuable data from textual input, a preprocessor is used to enhance the quality of the data. Next, pre-trained glove word embedding methods and term frequency-inverse document frequency (TF-IDF) feature weighting are used.