Information Technologies and Computer Engineering

The paper proposes an improvement of the method of extracting key words and phrases in the web-text. The following main stages of the formation of a plurality of key words and phrases are considered in order to find ways to increase the speed of indexing and refereeing web texts, to accurate source text, exclude stop words, cut off bases and endings from the text, the formation of key words and phrases from the source text. The proposed improvement is based on the use of the vocabulary of the subject area compiled by the expert. The dictionary is formed taking into account the frequency of repetitions of keywords and phrases in the web-text, will improve their relevancy. The comparison of the quality of the revealing keywords and phrases in the Ukrainian and English language web texts with the systems Expert Review, Open-Calais, Extractor, as well as the system based on the proposed method using the dictionary, recall, accuracy and F-measure. The analysis showed that the proposed advanced method for extracting keywords and phrases in Ukrainian and English web-texts will allow to reveal relevant words and word-received with an increase of their F-measures by 9.5%, and completeness and accuracy by 15%

References