Early Stage Researcher at Indra
Luca has a background in humanities and cognitive semiotics, with a keen interest in pragmatics and computational semantics. He joined Minsait, Indra to develop explainable AI techniques for the construction industry. He enjoys cocktail culture and backpacking.
Indra Soluciones Tecnologías de la Información SL (Indra)
Avda. Bruselas 35,
28108 – Alcobendas, Madrid, Spain
Follow
Explaining contracts and documentation of assets for construction companies and real estate agents.
· Main Supervisor: Mr. Juan Prieto Vivanco, Minsait- Indra Soluciones Tecnologías de la Información SL (Indra), jprietov@minsait.com
· Main Supervisor: Ms. Cristina Justo Suarez, Indra Soluciones Tecnologías de la Información SL (Indra)
· PhD Co-Supervisor: Prof. Dr Senén Barro Ameneiro, Centro Singular de Investigación en Tecnoloxías Intelixentes (CiTIUS)- Universidade de Santiago de Compostela (USC)
· PhD Co-Supervisor: Assoc. Prof. Dr. Pablo Gamallo Otero, Centro Singular de Investigación en Tecnoloxías Intelixentes (CiTIUS)- Universidade de Santiago de Compostela (USC)
· Inter-sectoral Secondment Supervisor: Dr. Mariët Theune, Human Media Interaction Research Group – Universiteit Twente (UTWENTE)
· Inter-sectoral Secondment Supervisor: Dr. Nava Tintarev, Department of Software Technology – Technische University of Delft (TU Delft)
PhD research topic
Objectives:
To develop new Explainable Artificial Intelligence (XAI) techniques for the construction industry. Construction and real estate companies manage thousands of unstructured texts in the form of contracts, certificates, deeds, legal notices, gazettes and other text documents. Manually reviewing and comparing these documents is exhausting, costly and time consuming. AI and NLP-based techniques such as text classification, information extraction, named entities recognition, entity linking, automatic summarization, topic extraction and semantic analysis may enable the iterative management and access to huge amounts of document collections in more intelligent and efficient manners.
However, the main business challenge rises from designing virtual intelligent agents ready to assist humans in the understanding and use of those techniques. Namely: analysing the state of the art solutions, analysing potential GDPR-related risks, estimating expected impact in the proposed scenarios and possible go-to-market actions, and communicating/explaining findings to humans in NL.
Use case:
Theoretical contributions will be validated in a real use case to meet the stakeholder needs.
Expected Results:
(1) Understanding of the state of the art in relevant aspects of NL generation/NLG for interactive XAI.
(2) Definition and implementation of a new methodology for analysing and explaining contracts and documentation of assets for construction companies and real estate agents.
(3) Validation of this methodology in a real use case.
