A libary is usually comprised of hardcopy collections, hardcopy collections are usually represented by MARC metadata. MARC metadata stores title, miltiple authors, subject, publisher, and several other data about the collection. to retrieve hardcopy metadata a library usually use a simple text matching algorithm. the weakness of text matching algorithm. the weakness of text matching algorithm is that this algorithm cannot sort its results based on relevance rherefore its usability is reduced. this study proposes the use of term expansion, vector based similarity measurement,and collections borrowing history to calculate each metadata record's relevance with user's query terms.