Similar Section Extraction for the Analysis of the Stream Data Structure

The present paper proposes a new algorithm for discovering similar sections between two time sequence data sets. The algorithm, called Partial Matching Discovery, or PMD, is based on Dynamic Programming. PMD realizes fast matching between arbitrary sections in reference stream data and input stream data and enables the extraction of similar sections in a synchronous manner with the input data. The algorithm can be used for the discovery of stream data structure. If similar sections, or the same sections, can be discovered in stream data, various types of applications, such as topic segmentation, CM parts identification, and structure similarity discovery of stream data, such as brain wave data, can be considered. In the present paper, the concept and the PMD algorithm are described in detail. In addition, experiments are conducted in order to evaluate the performance of the PMD algorithm in detecting similar sections in an English speech corpus The results show that PMD successfully detects same word sections. The proposed method can be applied to any type of stream data, including video imagery. Finally the possibility of the discovery of stream structure is discussed.. . .