Modeling the visualization and exploration of document collections with user and purpose-based adaptation

• Autores: Cristian Moral Martos
• Directores de la Tesis: Angélica de Antonio Jiménez (dir. tes.) , Xavier Ferré Grau (codir. tes.)
• Lectura: En la Universidad Politécnica de Madrid ( España ) en 2016
• Idioma: español
• Tribunal Calificador de la Tesis: Paloma Díaz Pérez (presid.) , Ricardo Imbert Paredes (secret.) , Angel Rodríguez Martínez de Bartolomé (voc.) , Matthieu Poyade (voc.) , Salvador Sánchez Alonso (voc.)
• Enlaces
  • Tesis en acceso abierto en: Archivo Digital UPM
• Resumen

  • Seeking information is an essential activity in the field of research. However, due to the huge amount of information that already exists, and that is increased every day, it is difficult to find the specific information that a researcher actually needs. Many software tools have appeared in the last years to facilitate information-seeking activities, but most of them are based on the classic “best-match” approach. However, a “one-size-fits-all” solution is limited in order to effectively fulfill the user’s information needs, especially in the research domain, where huge document collections are managed, and very specific information is usually required to achieve the user’s purpose.

    Instead, information seeking and exploration systems should effectively adapt to all the aspects that may influence the information-seeking process, like the user, the type of data being used, the context of the seeking activity, or the purpose the user aims to achieve. An essential precondition for adaptation is the existence of a clear model of this process that considers all the concepts that intervene, and how they interrelate. However, such a model is still missing.

    To fill this gap, in this research we have carried out a complete inductive qualitative study allowing us to comprehensively understand how the process is performed by researchers and which are the relevant concepts and relationships that intervene in it. The research has been contextualized in the computer science field to ensure that the possible variability in the information-seeking practices in different domains does not bias the obtained information.

    Based on the results of this qualitative study, we propose representing the concepts and relationships that emerged in the analysis of the process through a set of holistic and extensive conceptual models. In order to facilitate their understanding, a pictorial representation has been created using a standard, easy-to-understand and widely used representation language (UML). First of all, we propose a conceptual model of the information-seeking process, where high-level concepts, like document, task, or purpose, are not only present, but are also modeled in detail.

    On the other hand, due to the huge amount of documents that are usually involved in seeking activities, information visualization has become an essential aspect of the process, as it potentially can transmit information (like relationships) in a very intuitive and effective way. For this reason, the concepts and relationships specifically related to the information visualization activities have also been described through a conceptual model.

    Finally, the researcher’s characteristics and preferences that are relevant for information seeking and exploration are also described in a conceptual model.

    In all cases, the models have been designed to reflect the complexity of the process, and are also flexible enough to be easily modified or extended.

    Once the processes and actors have been fully modeled, we propose a fuzzy logic approach to allow a dynamic and adaptive calculation of the relevance of certain information elements (like authors or documents) depending on the user preferences and on the interrelationships that exist among them (as specified in the conceptual model).

    In order to point out the usefulness of the proposed solution, we present some of its practical applications. First of all, the models provide a complete framework that can be used to comprehensively analyze, describe and compare in detail existing information systems. Additionally, models can also be used to guide the design of new information systems and some prototypes developed with the model guidance are presented to prove the feasibility of such application of our proposal.