Discovering Explanatory Sentences in Legal Case Decisions Using Pre-trained Language Models

Legal texts routinely use concepts that are difficult to understand. Lawyers\nelaborate on the meaning of such concepts by, among other things, carefully\ninvestigating how have they been used in past. Finding text snippets that\nmention a particular concept in a useful way is tedious, time-consuming, and,\nhence, expensive. We assembled a data set of 26,959 sentences, coming from\nlegal case decisions, and labeled them in terms of their usefulness for\nexplaining selected legal concepts. Using the dataset we study the\neffectiveness of transformer-based models pre-trained on large language corpora\nto detect which of the sentences are useful. In light of models' predictions,\nwe analyze various linguistic properties of the explanatory sentences as well\nas their relationship to the legal concept that needs to be explained. We show\nthat the transformer-based models are capable of learning surprisingly\nsophisticated features and outperform the prior approaches to the task.\n