Abstract
Despite the relevance of contractual conflict in legal practice, there is yet to be a dataset which captures the type of issues and clauses that result in cases being brought before the courts. Such a dataset would be invaluable to a machine learning algorithm that seeks to predict whether new clauses are likely to cause conflict. In this study, we analyse a dataset based on half a million United Kingdom court decisions decided between 1709 and 2021, from which we extract 60,379 cases dealing with contracts. We characterise the language of this dataset using Latent Dirichlet Allocation to approximate legal topic modelling. We augment the data by plotting it with the court names and dates for each case, which allows for a racing bar chart visualisation. This is the first study of its kind to provide easy access to legal researchers on cases dealing with contracts in the United Kingdom.