A Hybrid Neurofuzzy System for Legal Text Comparison and Analysis
Résumé
The daily tasks of legal professionals are often hindered by the rapid pace at which new legislation is produced. Moreover, this legislation is typically presented in unstructured formats, unsuitable for automated computer processing. This leads to a vast amount of heterogeneous information being created in a disorganized way, resulting in information overload. To address this issue, we have developed a new model for comparing legal texts that integrates the latest advances in language processing through neural architectures with traditional fuzzy logic techniques. We have tested this model using the lawSentence200 benchmark dataset, and the initial results are promising.