A Kind of Index for Content-Based Music Information Retrieval and Theme Mining

摘要: Content-based music information retrieval and theme mining are two key problems in digital libraries, where “themes” mean the longest repeating patterns a piece of music. However, most data structures constructed for retrieving can not be efficiently used to mine themes pieces, vice versa. The suffix tree structure both functions, nevertheless its height is too large maintenance somewhat difficult. In this paper, kind index introduced, which adopts idea inverted files that N-gram. It retrieve as well themes. Based on several useful concepts, algorithm proposed. Also, implementations content-based presented. Experiments show correctness efficiency proposed algorithms.