Research

 People

 Publications

 Sponsors & Affiliates

 News

 Related Sites

 Offered Courses



New Metrics for Measuring the Effectiveness of Responses to Queries for Information Requests by Hyun Shin

Introduction
Most keyword-based information retrieval systems have limited facilities to accept and process a user’s intention. The user’s request is commonly provided by terms, phrases, or sentences. There is no accommodation for a system to collect and analyze the user’s intention. In order to analyze the user’s intention, we adopt user modeling scheme. The goal of user model is to accurately capture and represent a user’s intent. User modeling techniques have been exploited to help users, including analysts, improve their performance since the late 80s (Brajnik et al 1987). The successful user model should be an answer to improve retrieval performance and satisfies users simultaneously.

Philosophy for quantifying the generality
In the traditional information retrieval systems, index terms are used to index and retrieve documents. An index term is a keyword (or a group of related words) whose semantic reference serves as a mnemonic device for recalling the main themes of the document. Thus, an index term set is simply keywords which appears in the text of document in the collection. The index term set is attached to each sub collection that is built by ontology mechanism.
Within this philosophy, the degree of generality can be quantified by the amount of the index terms in the document that belong to specific word sets. The degree of generality can be quantified by the amount of the index terms in the document that belong to specific word sets. The specific word set is a sub set of the index term set that is not appeared in other sub collections (ontology node). A specific word set of each node consists of index terms that do not belong to the index terms set in other node. For example, a node has its own index term set . Then a specific term set is a set of index terms that are not belonging to index term sets in other nodes.

The Figure 1 illustrated a diagram of the degree of generality in document . In the figure, and are two different ontology nodes (or clusters). The document is an instance of . Therefore, the degree of generality for the document is presented by the dark gray portion. The specific word set of is presented by the light gray portion

Figure 1 A diagram presentation for the degree of generality for the document

The goal of this study is to introduce, define and quantify the degree of generality to find relevant information in response to a user’s intent. Since the traditional meaning of the relevance is not sufficient to satisfy the user’s intent to find the appropriate information, or to retrieve the relevant information, we need an additional criterion called “generality” to improve the retrieval results.


Home | Research | People | Publications | Sponsors & Affiliates | News | Offered Courses

© 2000-2013 Semantic Information Research Laboratory. All Rights Reserved.