Mathematical model of subsystem of document retrieval and ranking in information-retrieval systems

Author(s):  N.N. Kuchukova, FSAEI HE «North-Caucasus Federal University», Stavropol, Russia, knn.storage@yandex.ru

N.A. Vershkov, candidate of Sciences, FES-Agro, Stavropol, Russia, vernick61@yandex.ru

Issue:  Volume 45, №1

Rubric:  Infocommunication technologies

Annotation:  The article proposes the problem of fast and accurate information retrieval on the user's request in information retrieval systems. There is analysis of the existing models of searching and ranking documents and specification of the main performance parameters of information retrieval systems. It is revealed that modern search engines use a large number of parameters, according to which the completeness of the search and the relevance of the data obtained are estimated. This leads to an increase in computing and hardware costs. We offered the mathematical models of search and ranking documents in information retrieval systems, given the interdependence of terms in texts, the possibility of application of the apparatus of correlation to measure the similarity between a search query and document, and reduce computational costs. Also it is offered to use metric correlation as a single generalized indicator of information retrieval systems to establish the degree of similarity between the search query and documents, sort by relevance, with the exception of documents-duplicates from the search results

Keywords:  information retrieval system retrieval model, sorting, ranking, correlation, correlation coefficient

Full text (PDF):  Download

Downloads count:  435