User Centric Media. Second International ICST Conference, UCMedia 2010, Palma de Mallorca, Spain, September 1-3, 2010. Revised Selected Papers

Research Article

Multimodal Queries to Access Multimedia Information Sources: First Steps

Download
500 downloads
  • @INPROCEEDINGS{10.1007/978-3-642-35145-7_5,
        author={\^{A}ngel Mart\^{\i}nez and Sara Lana Serrano and Jos\^{e} Mart\^{\i}nez-Fern\^{a}ndez and Paloma Mart\^{\i}nez},
        title={Multimodal Queries to Access Multimedia Information Sources: First Steps},
        proceedings={User Centric Media. Second International ICST Conference, UCMedia 2010, Palma de Mallorca, Spain, September 1-3, 2010. Revised Selected Papers},
        proceedings_a={UCMEDIA},
        year={2012},
        month={12},
        keywords={multimodal queries search engine query by example results lists combination multimedia retrieval relevance feedback},
        doi={10.1007/978-3-642-35145-7_5}
    }
    
  • Ángel Martínez
    Sara Lana Serrano
    José Martínez-Fernández
    Paloma Martínez
    Year: 2012
    Multimodal Queries to Access Multimedia Information Sources: First Steps
    UCMEDIA
    Springer
    DOI: 10.1007/978-3-642-35145-7_5
Ángel Martínez1,*, Sara Lana Serrano1,*, José Martínez-Fernández,*, Paloma Martínez2,*
  • 1: DAEDALUS, Data, Decisions And Language, S.A.
  • 2: Universidad Carlos III de Madrid
*Contact email: amartinez@daedalus.es, slana@daedalus.es, jmartinez@daedalus.es, paloma.martinez@uc3m.es

Abstract

This position paper deals with queries beyond text, mixing several multimedia contents: audio, video, image and text. Search approaches combining some of these formats have been studied, including techniques in situations where only one format is considered. It is worth mentioning that most of these research works do not deal with text content. A new approach to allow users introducing multimodal queries and exploring multimedia repositories is proposed. For this purpose, different ranked result lists must be combined to produce the final results shown for a given query. The main goal of this proposal is to reduce the semantic gap between low level features and high level concepts in multimedia contents. The use of qualitative data giving more relevance to text content along with machine learning methods to combine results of monomodal retrieval systems is proposed. Although it is too soon to show experimentation results, a prototype implementing the approach is under development and evaluation.