Measuring Multimedia Content Proximity via Artificial Intelligence Methods

The paper describes existing and novel approaches to word sequence similarity measuring which uses image processing procedures to reveal proximity relations between words and their meaning. This is implemented due to the fact the each word sequence can be associated with a particular pictorial domain, and thus any textual information can be potentially presented via an image series. Theoretical results of rule based approach, genetic algorithms and neural networks application are analyzed in terms of solving the problem under study. The proposed information processing technology may be useful for automatic translation applications, known as CAT-tools by now, search engine optimization, etc. The proposed content analysis techniques are designed to boost translation efficiency in particular. Romano-germanic and east-slavonic language pairs have been used as an example of practical application. It aids translation experts in finding the best solution for a specific domain that he (or she) is usually unfamiliar with. Thus, eliminating human errors and decreasing time needed for contextual search and deep examining of the topic.