Practical evaluation of textual fuzzy similarity as a tool for information retrieval

Abstract

This paper presents a practical evaluation of a document retrieval method based on a certain textual fuzzy similarity measure. The similarity measure was originally introduced in [1] - cf. also [2], and later used in Internet-related applications [3,4]. Three textual databases of diverse level of freedom in the content of documents are used for experiments in the search. In other words, the relation of the documents within each group to the chosen topic is (according to the evaluating person) strong, average, and random. The results of the search coincide with intuition and confirm the expectation that methods based on similarity are advantageous as long as the database contains documents of a relatively well-defined topic.