Exploring Web Pages' Semantics and a Subject Hierarchy for Supporting Personalized Web Searches

Loading...

Flash Player 9 (or above) is needed to view presentations.
We have detected that you do not have it on your computer. To install it, go here.

0 comments

Post a comment

    Post a comment
    Embed Video
    Edit your comment Cancel

    1 Favorite

    Exploring Web Pages' Semantics and a Subject Hierarchy for Supporting Personalized Web Searches - Presentation Transcript

    1. Exploring Web Pages’ Semantics and a Subject Hierarchy for Supporting Personalized Web Searches International Conference on Multidisciplinary Information Sciences and Technologies (InSciT) October 25-28, 2006 Merida, SPAIN Sofia Stamou, Panagiotis Kapros, Dimitris Christodoulakis Computer Engineering and Informatics Department Patras University, Greece {stamou, kapros, dxri}@ceid.upatras.gr
      • Personalized Web searching is the process of adapting the information that is relevant to a search query to the particular needs / interests of a user or groups of users .
      Introduction Exploring Pages’ Semantics and a Subject Hierarchy for Supporting Personalized Web Searches InSciT 2006
    2. Web personalization Advantages Exploring Pages’ Semantics and a Subject Hierarchy for Supporting Personalized Web Searches InSciT 2006
      • Users obtain only relevant information
      • Increased accuracy of search results
      • Improved user search experience
    3. Web personalization Challenges Exploring Pages’ Semantics and a Subject Hierarchy for Supporting Personalized Web Searches InSciT 2006
      • Automatically creating and maintaining accurate user profiles within large-scale Web Search Engines requires significant effort
    4. The main difficulties Exploring Pages’ Semantics and a Subject Hierarchy for Supporting Personalized Web Searches InSciT 2006
      • Build accurate profiles for the users’ interests
      • Modify the profiles when necessary (e.g. when user interests change over time)
      • Adjust search results to fit user interests
      • Greatest Difficulty …
      • Users’ reluctance to give information about their search interests!
    5. Common approaches to personalization Exploring Pages’ Semantics and a Subject Hierarchy for Supporting Personalized Web Searches InSciT 2006
      • Explicit user feedback : adopted my most commercial systems. Users define their interests.
      • Implicit user feedback : monitor the users search behavior and infer their interests based on their past click history
      • Content-based implicit user feedback : learn the user interests based on the semantic content of the previously visited pages
    6. Our approach
      • We employ a subject hierarchy for annotating user issued queries with search topic intentions
      • Identified user interests are considered in the search process as user profiles
      • Retrieved results contain pages that satisfy the topical categories used to describe the user interests
      Exploring Pages’ Semantics and a Subject Hierarchy for Supporting Personalized Web Searches InSciT 2006
    7. Outline…
      • We classify web pages in the hierarchy’s topics
      • We map the terms in the issued queries to the hierarchy’s topics
      • Retrieved results contain pages that satisfy the topical categories used to describe the user queries
      Exploring Pages’ Semantics and a Subject Hierarchy for Supporting Personalized Web Searches InSciT 2006
    8. Getting started on pages’ classification
      • Download Web pages
      • HTML parse, tokenize, POS-tag, lemmatize
      • Generate shingles (Broder et al., 1997)
      • Find thematic words inside the pages’ shingles, using the lexical chaining technique (Barzilay and Elhadad, 1997)
      Exploring Pages’ Semantics and a Subject Hierarchy for Supporting Personalized Web Searches InSciT 2006
    9. Reducing web pages’ to lexical chains
      • Select a set of candidate terms from every shingle
      • For each candidate term find an appropriate chain relying on the type of WordNet links that connect the candidate term to the terms already stored in existing lexical chains
      • If it is found, insert the term in the chain and update accordingly
      Exploring Pages’ Semantics and a Subject Hierarchy for Supporting Personalized Web Searches InSciT 2006
    10. Categorizing pages in the hierarchy
      • Map pages’ thematic words to the hierarchy’s concepts
      • Following hypernymic links of the hierarchy’s matching nodes, reach to topic concepts
      • Compute a Relatedness Score ( RScore ) of every page to each of the hierarchy’s matching topics
      Exploring Pages’ Semantics and a Subject Hierarchy for Supporting Personalized Web Searches InSciT 2006
      • Classify a page to the topic of max RScore(i,k) ≥ T
    11. Mapping query terms to the hierarchy
      • Map all queries issued by a user to the hierarchy’s concepts
      • Disambiguate query terms based on their semantic relations in WordNet
      • Compute the Correlation Score between query terms following the formula:
      Exploring Pages’ Semantics and a Subject Hierarchy for Supporting Personalized Web Searches InSciT 2006
      • Select the WordNet senses that maximize the Correlation score for query terms disambiguation
      • Pick the topical category of the WordNet matching nodes for describing the query topics
    12. Evaluation of personalization accuracy
      • Experimental setup
        • 17 experienced Web users
        • 10 self-defined queries/user describing a specified topic of interest
        • total set of 170 queries
        • Issue self-selected queries twice: with and without activating the personalization search
        • Review top 10 pages retrieved for each of the queries, in each of the search modes (plain and personalized) and evaluate relevance on a 5-point scale
      Exploring Pages’ Semantics and a Subject Hierarchy for Supporting Personalized Web Searches InSciT 2006
    13. Personalization performance Exploring Pages’ Semantics and a Subject Hierarchy for Supporting Personalized Web Searches InSciT 2006
    14. Conclusions Exploring Pages’ Semantics and a Subject Hierarchy for Supporting Personalized Web Searches InSciT 2006
      • Relevance of personalized results is analogous to the hierarchy’s richness
      • Hierarchy-based personalization has a notable potential in improving the user search experience
      • User profiles can be built automatically with the use of a subject hierarchy
      • Dynamic profile construction and updates
    15. Thank You!! Exploring Pages’ Semantics and a Subject Hierarchy for Supporting Personalized Web Searches InSciT 2006

    + inscit2006inscit2006, 3 years ago

    custom

    1189 views, 1 favs, 0 embeds more stats

    Sofia Stamou, Panagiotis Kapros and Dimitris Christ more

    More info about this document

    © All Rights Reserved

    Go to text version

    • Total Views 1189
      • 1189 on SlideShare
      • 0 from embeds
    • Comments 0
    • Favorites 1
    • Downloads 0
    Most viewed embeds

    more

    All embeds

    less

    Flagged as inappropriate Flag as inappropriate
    Flag as inappropriate

    Select your reason for flagging this presentation as inappropriate. If needed, use the feedback form to let us know more details.

    Cancel
    File a copyright complaint
    Having problems? Go to our helpdesk?

    Categories