Measures of Concentration and Distribution of Knowledge Contained in Documents

摘要

An important parameter characterizing a document is the degree of knowledge concentration and distribution contained in it. This work describes and illustrates through examples the intuition that is connected with these notions. Next, measures of knowledge concentration and distribution are defined, and their characteristics are analyzed. To calculate concentration and distribution one uses the description of the contains of documents with the help of terms and their weights. Measures of concentration and distribution can be used in information retrieval systems as an additional parameter which helps the user to evaluate document relevance. Discussing the direction of further research we called attention to the necessity and possibility of solving the problem of term dependence.