Методы поиска ближайших соседей в задаче анализа графического образа структурированного документа

Abstract

In some tasks of artificial intelligence, the problem arises of searching among a multitude of information objects most similar to this material. For example, the ability to recognize text, information retrieval, data compression, classification and clustering, building a database of images, video documents. In order to build an effective search method, it is necessary to take into account the peculiarities of this task. For example, strings must be processed in one way, and vectors in another. Therefore, the algorithms used will depend on how well it fits the description.