Financial decision-making process based on unstructured data sources and domain ontologies

• Autores: Mateusz Karol Radzimski
• Directores de la Tesis: Ángel García Crespo (dir. tes.), José Luis López Cuadrado (codir. tes.)
• Lectura: En la Universidad Carlos III de Madrid ( España ) en 2017
• Idioma: español
• Tribunal Calificador de la Tesis: Antonio Bibiloni Coll (presid.), Israel González Carrasco (secret.), María Inmaculada Puebla Sanchez (voc.)
• Materias:
  • Matemáticas
    • Ciencia de los ordenadores
      • Inteligencia artificial
• Enlaces
  • Tesis en acceso abierto en: e-Archivo
• Resumen

  • Nowadays a great number of financial decisions arrive from watching the information stream, selecting relevant data, analysing it and acting accordingly. With the increasing global competition, the need for swift data analysis, high accuracy and quality becomes a must.

    For the majority of financial analysts, the main source for information is in the form of structured data. Such data can be easily processed and acted upon. However, there are vast amounts of knowledge that still can not be easily digested by computers, but have a great importance in our everyday life. For instance, (i) news are describing events and changes to the state of the world, (ii) columnists' opinions are providing arguments that are shaping our thoughts or (iii) experts' conclusions are influencing people decisions.

    This thesis main objective is to employ unstructured data in the financial decision-making process, with the support of ontologies as the main backbone for knowledge representation. The whole financial-making process is contextualised in the scope of the Spanish market, where the main source of data is news and company disclosures published in the Spanish language.

    The main contribution of this thesis is the creation of the \emph{Decision Support System} (DSS) that follows a novel approach to incorporate unstructured data and domain (financial) ontologies into the automated financial decision-making process. Our approach employs Natural Language Processing (NLP) as means for extracting relevant information from unstructured sources. Moreover, semantics is applied thoroughly, not only in the process of information extraction but also in the knowledge modelling and the decision support.