IOSR Journal of Computer Engineering (IOSR-JCE) e-ISSN: 2278-0661,p-ISSN: 2278-8727, Volume 17, Issue 5, Ver. V (Sep. – Oct. 2015), PP 26-29 www.iosrjournals.org DOI: 10.9790/0661-17552629 www.iosrjournals.org 26 | Page Music Data Organization Using Heuristic Hierarchical Agglomerative Co-Clustering N. V. Keerthana 1 , S. P. Yazhini 2 1 Department of information Technology,Kongu Engineering college/Anna University,India 2 Department of information Technology,Kongu Engineering college/Anna University,India Abstract: In Data mining, Information Retrieval (IR) is the study of significant recent interest. In IR the Music Information Retrieval (MIR) is unitary of the challenging problems. At MIR, the utilization of Tags, Styles and Mood labels (T/S/M) can be extracted from some music websites. A typical problem is how to understand the relationship between the T/S/M. Co-clustering is the combination of two different types of data simultaneously. Hierarchical Co-clustering (HCC) extracts the acoustic features of music to find the similarity among the artist. In this paper, we systematically analyze the application of Heuristic Hierarchical Agglomerative Co-clustering (HHACC) method for music data organization. We demonstrate that this HHACC method achieves better performance than other clustering methods. Keywords: Hierarchical Co-clustering, HACC, HHACC, T/S/M. I. Introduction The music information retrieval research is the interdisciplinary science of retrieving information from music. It has a number of applications concerned with classification, clustering, indexing and searching in musical database. Traditional musical classification approaches usually assume the each piece of music has a unique style and they make use of the music content to construct the classifier. This classifier is identically used for classifying each piece of music information into it unique style. The music information is particularly differentiated as Tags, Style and Mood labels. The style and mood labels provide special features to define the similarity between the artist and the music tags. Generally the tags and labels are assigned to individual pitch, not to the artist. So by sampling the pitch of an artist, one is able to know tag, style and mood labels of the artist. Hierarchical clustering is the method of analyzing the cluster, which is used to build a hierarchy of clusters. In hierarchical clustering the series of data partitions takes place, which may range from a single cluster containing all objects to the number of clusters each containing an exclusive target. Hierarchical clustering offers an extended description of document browsing [2]. There are two types of hierarchical clustering, one is divisive method and the other is agglomerative method. The former method divides the data set into smaller groups iteratively and the later one is the reverse process of divisive approach. Co-clustering or two-mode clustering is a data mining technique clustering of multiple data simultaneously. After analyzing with the proposed hierarchical divisive co-clustering (HDCC) method and hierarchical agglomerative co-clustering (HACC), we present a fictitious method, heuristic hierarchical agglomerative co-clustering (HACC) method. The divisive HDCC combines K-means and similar value decomposition (SVD) [1]. The HACC method starts with a single cluster and then iteratively merges two nearby clusters into one cluster until all the points are merged into a single cluster. In the case of HHACC, at each step of merging procedure, HHACC can merge a subset of the T/S/M labels and the subset of the artist. Thereby it needs to construct double-hierarchical for both artist and T/S/M. HACC merges the artist and T/S/M into a single group at the earliest possible stage [4]. Our finish is that search clusters with two types of data will be employed for better retrieval when both types of data designated in a query, e.g., given a query with an artist and one of its T/S/M, one can probably retrieve them together from a creative person-tag hierarchy, while with the query composed of an artist and style, one can retrieve simultaneously from an artist-style hierarchy. In this paper, we demonstrate that such mixed-data-type hierarchical cluster can generate HCC and empirically better cluster generated by concurrent usage of two data types. Our contributions in this paper are: 1) we develop a fictional hierarchical agglomerative co-clustering method to organize a music data. 2) We analyze a demonstration to show that HHACC have the capacity of providing reasonable artist similarity qualification measures. II. Existing Work Hierarchical clustering is the operation of generating the clusters that take in a predetermine ordering in the form of tree like cluster structure of partitions. Hierarchical clustering algorithms organize input data either bottom up (agglomerative) or top down (divisive) [3]. Generally, hierarchical agglomerative clustering is more practiced than the hierarchical divisive clustering. Co-clustering is the clustering of more than one data