Concept based interactive retrieval for social environment

  • Authors:

  • Tijana Janjusevic;Qianni Zhang;Krishna Chandramouli;Ebroul Izquierdo

  • Affiliations:

  • Queen Mary, University of London, London, United Kingdom;Queen Mary, University of London, London, United Kingdom;Queen Mary, University of London, London, United Kingdom;Queen Mary, University of London, London, United Kingdom

  • Venue:
  • Proceedings of the 2010 ACM workshop on Social, adaptive and personalized multimedia interaction and access
  • Year:
  • 2010

Quantified Score

Hi-index 0.01

Visualization

Abstract

Following the recent developments in social networking, there is an emerging interest to share experiences online with social peers through multimedia data. Consequently, exponential amount of multimedia information has been generated by everyday users and shared among social peers. As opposed to conventional digital archives, the user generated content archive does not confine to one particular domain and therefore semantic indexing of the content requires the creation of large number of training samples for each semantic query concept. Addressing this problem, we present an interactive multi-concept based browsing and retrieval framework using which users can construct high-level semantic queries based on mid-level primitive features. The proposed framework integrates innovative visualisation methodology developed for browsing, navigating and retrieving information from multimedia database. The framework is user centric and supports interactive formulation of high-level semantic queries for content retrieval using available content annotation. The performance of the proposed framework is evaluated using annotation based on automatic algorithms against Support Vector Machines, Multi-feature classification and particle swarm optimisation based relevance feedback techniques.