Slide KWIC: Snippet Generation for Browsing Slides Based on Conceptual Relationship and Presentational Structure

Abstract

Currently, several presentation files such as Slide Share and MPMeister are shared by many universities over the Web. These files are useful and valuable to students. However, such files have to be retrieved for self-learning purposes, and there is still a lack of support for self-learners browsing through slides containing information that might be irrelevant to their query. We describe a slide retrieval method involving the snippet generation of target slides, and we discuss how to present the retrieval results to users by considering what portions of the slides are relevant to a user query, on the basis of the relationships between slides. This method is based on (1) the keyword conceptual structure of the conceptual relationship that implicitly exists between keywords that are extracted from the slide text of the presentation content by Word Net and (2) the presentational structure of the indent levels in the slides. With our novel snippet-generation method, users can easily determine which slides to learn by browsing the relevant portions of the desired slides in the presentation and by focusing on portions from either detailed or generalized slides. We experimentally confirmed that our method enables the users to browse snippets of the retrieved slides efficiently and effectively.