Applying Human Language Technology to the Law

This paper relates preliminary results towards assessing the possible problems related to the reuse of existing thesauri for the search and retrieval of legal information. In particular, it focuses on the study of the Code of Federal Regulations and its related thesaurus, the Thesaurus of Indexing Terms. The relationship between the content of a thesaurus and the content of the objects which represents may be of different nature depending on the thesaurus’ production process. Here we would like to study the related nature of the content of the Code of Federal Regulations and the Thesaurus of Indexing Terms. In order to assess the extent to which the indexing terms and the content of the CFR headings are equally representative of the content of a CFR title, we use a scaling method that will deliver the position of each CFR title with respect to a one-dimensional space, in particular, the document scaling algorithm wordfish. Results show that , although there is in general a strong correlation between the scores received by the CFR titles in both experiments, the existence of mismatches suggests that part of corpus of indexing terms is not representative of the actual content of the CFR titles based on the analysis of their headings. This, in turn, may suggest that the direct reuse of an existing thesaurus for textual search and retrieval could, in some cases, lead to inefficient results.