Topic Modelling of Czech Supreme Court Decisions: Evaluation of Legal Experts
| Authors | |
|---|---|
| Year of publication | 2021 |
| Type | Appeared in Conference without Proceedings |
| MU Faculty or unit | |
| Citation | |
| Attached files | |
| Description | Topic modelling is set of common state-of-the-art methods used for more precise legal information retrieval. However, the relevance of topics they assign to the documents can significantly differ from human evaluation. In this paper proposal we describe evaluation of legal experts of two of the topic modelling methods we used on the dataset of the Czech Supreme Court decisions dataset. We use the latent Dirichlet allocation and non-negative matrix factorization for obtaining of topics of court decisions. Subsequently, we design evaluation of legal experts experiment to evaluate the relevance of these topics. We discuss the strengths and weaknesses of these methods regarding the evaluation results and suggest their improvement to gain more useful topics of court decisions. |
| Related projects: |