Most Significant Research Contributions

My research contribution is centered on the application of theoretical linguistics to specific real-world problems, in particular to automated text-based question answering and summarisation. Since 2009 I have focused on medical text processing.

I joined the University of Zurich and became the principal researcher in the ExtrAns and WebExtrAns projects (from 1996 to 2001). Both projects were based on the development of answer extraction systems. Answer extraction systems locate those sentences in the source text that contain the answer posed by the user. The outcome of the project was a working system that handles questions about 500 Linux/Unix documentation documents ("manpages"). I was the principal designer of the overall system and the integration of all modules. I was also the principal contributor to the design of the logical forms and the question-answering method that used the logical forms. The success of ExtrAns and WebExtrAns is evident from the fact that the system is cited as a pioneering question-answering system, an example of a question answering system of technical domains, and an example of the use of logical forms for question answering.

In Macquarie University I established the AnswerFinder project. AnswerFinder is a question answering system that combines the use of logical information (inspired from ExtrAns and WebExtrAns), state-of-the-art approaches in question answering, and innovative graph-based machine learning methods to find the exact answer to the user question. AnswerFinder has participated in the question answering track of the Text REtrieval conference (TREC), the main international forum for the evaluation of question-answering systems, between 2003 and 2006. AnswerFinder is in fact the only Australian-based question answering system that has participated in the question answering track of TREC.

Since 2009 I have led various projects related to the development and application of text-processing technologies that help the medical doctor find and appraise clinical evidence found in the vast resources of medical publications. I have gathered a summarisation corpus sourced from the Journal of Family Practice, and used the corpus to develop and evaluate techniques for text summarisation, clustering, keyword extraction, and appraisal of the medical evidence.

More recently, I have been working on the use of Large Language Models and Agentic design to develop task-oriented conversational agents. I have also been working on the analysis of social media posts for mental health.

CV: http://comp.mq.edu.au/~diego/cv.html