The World Wide Web has provided the facility of bringing information to the fingertips of its users. Since most of the document available on the web are machine-readable but not machine-under tandable, ensuring the retrieval of relevant information continues to be a difficult task. In the traditional text representation approach, high frequency keywords are used as term representative of text. However, the main drawback in this approach are lack of direct relationship between word frequency and it importance, and the effect of the word ambiguities. Considering these shortcomings of the keyword-based method, this paper present a phrase-based text representation approach that uses rule-based Natural Language Processing (NLP) techniques. Extraction of key-phrases from text documents is based on a process of partial parsing. By making the indexing term more meaningful through reduction of the ambiguity in word considered in isolation, improvement in retrieval effectivenes is sought to be achieved.