Approach to Extract Keywords and Keyphrases of Text Resources and Documents in the Kazakh Language

摘要

In this paper authors propose a hybrid approach for extracting keywords and keyphrases of text resources and documents in Kazakh. Direct application of the statistical method tf-idf is not the optimal solution to the question of extracting keywords and phrases in the Kazakh language, since the Kazakh language is an agglutinative type of language. The authors developed and used the stemming algorithm in the pre-processing process taking into account the grammatical features of the Kazakh language. In the extraction, we also take into account the syntactic feature of the words or phrases using the morphological analyzer of the Kazakh language. During extraction, the restrictions indicated by the authors are observed as well, as not all words may be key words. When choosing keywords or a phrase, their features are considered (for example, some words that are a numeral name in combination with a noun are selected). The extraction of keywords and phrases specifically for the Kazakh language is an urgent task in classification, clustering, abstracting the text, and searching the information. The results of the research indicate that the presented approach is the best solution on extracting keywords and phrases from texts in the Kazakh language.