An integrated management of multi-media, including voice, video, and document data would become essential in a next generation communication systems for computers and information networks. As the natural language is considered definitely advantageous to represent such multi-media data, an intelligent communication system wherein a character recognizer, translation device enabling communications between different languages, speech-rule synthesizer and an image retrieval system by which this system is systematically integrated, and a natural language processing algorithm for managing such system are described here in addition to reports on the results of evaluation and problems arisen therefrom when this system is practically applied.