Method for Indexing and Identifying Multimedia Documents

Abstract

The method for indexing multimedia documents comprises the following steps: (a) identification and extraction, for each document, of terms ti comprising vectors characterizing the properties of the document; (b) storage of terms ti in a term base comprising P terms; (c) determination of a maximum number N of desired concepts including the most relevant terms ti; (d) calculation of the matrix T of distances between the terms ti of the term base, (e) decomposition of the set P of terms ti of the term base into N parts Pj (1=j=N) such that P = P1 UP2 UPj UPN, each part Pj comprising a set of terms tij and being represented by a concept cj, terms ti being divided in such a manner that the furthermost terms are located in distinct parts Pl, Pm and the nearest terms are located in the same part Pl; (f) structuring of the dictionary of concepts and (g) construction of a prints base consisting of the set of concepts ci representing the terms ti of said documents, each document being associated with a print which is proper thereto.