Harmony search (HS) is a meta-heuristic algorithm mimicking the improvisation process of musicians. This paper arranges the basic structure of the HS algorithm and customizes the algorithm for clustering optimization problems. We propose novel clustering algorithm based on Harmony Search (HS) optimization method that deals with document clustering. By modeling Retrieval as an optimization problem, first, we propose a pure HS based clustering algorithm, then we hybridize Harmony clustering with TABU to achieve better retrieval. Experimental results reveal that the proposed algorithms can find better results when compared to HS and TABU methods and the quality of results is comparable. Then Relevance feedback mechanisms such as term feedback, cluster feedback and term-cluster feedback are used to further improve the retrieved results. Finally we propose abstract generation system based on rhetorical structure for generating abstract for the retrieved documents.