Generating a reference set

Abstract

Information retrieval is the science of returning data from a corpus (a large collection of documents) matching the user's informational need. It identifies the data (originally in document form) by matching the terms in the query with terms contained in the documents of the collection. Representing documents and queries for effective retrieval is best accomplished by defining a model. Among the various models, the one most frequently used is Salton's Vector Space Model [9]. In this model, documents and queries are represented as vectors. The similarity between the query and a document is found by using a similarity measure (e.g., cosine). Extensible Markup Language (XML) is a simple, flexible text format derived from Standard Generalized Markup Language (SGML) [3], designed to meet the challenges of large-scale electronic publishing, XML plays an important role in the exchange of a wide variety of data on the Web and elsewhere. INEX (The Initiative for evaluation of XML retrieval) [1] sponsors a competition that promotes the development of XML-based retrieval. It provides a Wikipedia collection in the form of XML files and each of these XML files are well defined and documented. We are interested in building a reference run for a given set of queries without first performing a separate retrieval run to produce it. To that end, we perform some basic experiments which are based on the terminal nodes of the document set. The experiments depend upon the content of these nodes. Analysis of these early results, conclusions, and suggestions for future research are included.