Università degli studi di Udine
Dipartimento di Matematica e Informatica


Artificial Intelligence
Laboratory


Intelligent Information Retrieval Group


Goals
Results
Principal Investigators
Projects
Events
Research Partnership
Prototype


Goals

Intelligent Information Retrieval (IIR) is the intersection of the areas of Information Retrieval (IR) and Artificial Intelligence (AI). It aims at improving the performance of an IR system by means of AI techniques. In this scenario, the following activities take place in our AI laboratory:

FIRE Implementation
The FIRE prototype, whose main goal is to emulate some of the functions of a human intermediary when accessing a boolean IR system, has been developed and is currently being improved. FIRE capabilities are accomplished through AI techniques: explicit reasoning and representation of knowledge about the intermediary skills and about the subject domain are used by a knowledge-based module included in FIRE. Thus FIRE is an intelligent interface to an IR system, a particular instance of an IIR system.
FIRE evaluation
The FIRE prototype has recently been evaluated, in collaboration with researchers at the Department of Psychology of the University of Trieste. Evaluating intelligent interfaces to IR systems is a complex activity and evaluation methodologies that go beyond the classical precision and recall figures are not well established. In this line of research we propose an approach to the evaluation of an intelligent interface that covers also the user-system interaction and measures user's satisfaction. More specifically, we exploited an experiment that evaluates: (i) the added value of the semi-automatic query reformulation implemented in FIRE; (ii) the importance of technical, terminological, and strategic supports and (iii) the best way to provide them. The interpretation of results lead to guidelines for the design of user interfaces to IR systems and to some observations on the evaluation issue.
Cognitive-epistemic modelling of the IR activity
The lackness of a formal account is probably one of the most evident of the shortcomings of IR: concepts like information, information need, and relevance are neither well understood nor formally defined. A cognitive framework, that permits to analyze these three central concepts of the IR scenario, has been proposed and is currently being revised.
Understanding the nature of relevance
Notwithstanding its importance, and the huge amounts of research on this topic in the past, relevance is not yet a well understood concept, also because of an inconsistently used terminology. This issue can be clarified by a classification of the various kinds of relevance. It is shown that: (i) there are many kinds of relevance, not just one; (ii) these kinds can be classified in a formally defined four dimensional space; (iii) such classification helps us to understand the nature of relevance and relevance judgement and can be used for handling the huge literature on this subject and for analyzing the design and evaluation of IR systems.

Results

MAIN PUBLICATIONS:

Principal Investigators


Projects


Events


Research Partnership


Prototype


Previous page AI Lab