The advent of e-commerce and corporate intranets has led to the growth of organizational repositories containing large, fragmented, and unstructured document collections. Though it is difficult to retrieve relevant documents from such collections, it is relatively less cumbersome to define categories broadly classifying the information contained in the collection. Such categories lend value to the information contained in the collection. This research addresses the issue of improving retrieval accuracy of search engines that retrieve documents from organizational repositories using a value based approach. We test an evolutionary algorithm approach on a document collection. The precision of the search algorithm improved from 40% in generation 1 of the algorithm to nearly 90% in generation 10,000.