Development of a Search Engine for Corpora Resource

The thesis “Development of a Search Engine for Corpora Resource” describes unique study about behavior of users during the search session, which aims to find new methods to improve linguistic search. The main purpose of the work is to find additional ways to speed up the search process by improving the way data is stored. So, the author puts forward two hypotheses related to the behavior of the user in the corpora search resource. The first hypothesis is related to the indexing of texts and the idea of storing certain parts of morphological patterns separately. It is assumed that if some morphological patterns are sufficiently frequency-based, it makes sense to index them separately, and subsequently call out for such a separate index. The second hypothesis relates directly to user behavior within a single session. The author claims that by conducting a study using the case, the user repeats some of his queries more than once, and therefore additional caching of results for the time of this search session will speed up the search. The study is conducted on the materials of the largest Russian-language corpora, such as NKRYA, Leeds, GIKRYA and others. All the scripts for processing the search logs of the mentioned buildings were written by the author of the work in the programming language Python.