THE DETERMINATION METHOD FOR CONTEXTUAL MEANINGS OF WORDS AND DOCUMENTS

Современные информационные технологии и IT-образование

Elizaveta A. Dorenskaya¹  Yuri A. Semenov¹ 

[1] Institute for Theoretical and Experimental Physics named by A.I. Alikhanov of National Research Centre «Kurchatov Institute», Moscow, Russia;

关键词: The problem of context recognition;    contextual meaning;    machine analysis;    semantic network;    tree of semantic links;    artificial intelligence;    word characteristics;    Monte Carlo method;   

DOI  :  10.25559/SITITO.14.201804.896-902

来源: DOAJ

【 摘 要 】

Problems and methods are considered for program context recognition of words and text documents. Survey of existent text processing methods is provided, simple numeric algorithm is given for determination of words and documents context with a help of semantic net, having a form of tree type graph. Semantic net structure is described in detail. Given semantic net is needed to fix basic word W1 context by means of words-meaning W2 coupled with it. Words W2 represent possible W1 context meanings. For every word W2 correspond some words-characteristics W3. At the context calculation the distances between words W2 and W3 are taken into account. The distances are measured in words between. Every word W3 has metrics, accordingto the concept proximity to W2. There is a table of words W1,W2 and W3 with their metrics values. At context document analyses there was taken into account case or number words variations. Simple formula for context calculation is presented. Method of results proofing with a help of Chebyshev inequality is also provided. The context analyses method was checked by Monte-Carlo simulations. Tables of investigation results are provided and some recommendation for algorithm parameters tuning and optimization are also given. The analyses showed that proposed method is quite effective for context estimation at text analyses, and for any systems, where one needs computer recognition of context.

【 授权许可】

Unknown