The InnovationQ application will be updated on Sunday, May 31st from 10am-noon ET. You may experience brief service interruptions during that time.
Browse Prior Art Database

Interactive Visualization of Ranking versus Relevancy

IP.com Disclosure Number: IPCOM000013322D
Original Publication Date: 1999-Oct-01
Included in the Prior Art Database: 2003-Jun-18

Publishing Venue



Main idea of invention Enhance any exiting information retrieval system with an automatic graph generation tool which plots the ranking versus the relevancy of retrieved data. Our invention can enhance any existing information retrieval system (which computes quantitative measures of the relevancy of retrieved and ranked data) with an automatic graph generation tool which plots the ranking versus t relevancy of retrieved data. In the default setting, the system draws a graph, starting from the data with the highest relevancy down to the lowest in a monotonically decreasing manner (for as many as 1000 of the documents the highest relevancy with respect to an input query). An example of a screen image of our invention is shown be Variations/User Options Some variations of this basic plot are: (1) graphing the relevancy as a function of ranking starting from the n-th ranked document; (2) allowing a user to view selected portions of the graph in greater detail (either through activation of the a mouse or pen) or typing in the rankings of the documents (e.g., ``from document ranked #75 to #125"); (3) embedding interactive links in the graphs, e.g., a mouse- activated operation which allows users to call up the titles of the top $N$ ranked documents by clicking on the corresponding section of the horizontal axis o (4) graphing the change in the relevancy in going from the i-th ranked to the (i+1)-st ranked data as a functi (this option will show users if/when a sudden, relatively large change in the relevancy occurs); (5) allowing users to select the number of singular values and singular vectors used to reduce the dimensionalit ranking and retrieval computations (for retrieval systems based on modeling document, keywords and queries vectors, then using a reduced dimensional subspace to speed up retrieval and reduce noise, e.g., Latent Seman Indexing (Deerwester et al. 1989), (Deerwester et al. 1990)); and (6) plotting results from several different results from query and retrieval on the same graph.