Work Part 4: Integration

Work Part 4 - ``Integration'' is organized in the following way.

A short description of WP4 as a whole follows.

WP4 - Objectives

The objective of WP4 ``Integration'', and main objective of the FERMI project itself, is to design the MIRLOG logic, a logic that will encompass in a unifying view the theories developed within Work Parts 1 to 3. The MIRLOG logic should allow the representation and reasoning about multiple views of documents and queries, such as those relating to structure, layout and content. It should also allow the representation of domain knowledge relating, i.e. of knowledge allowing the interpretation ``in-context'' of the previously mentioned types of knowledge. It should also allow the representation of the various kinds of uncertainty (i.e. deviance from the yes-no case) that inherently affect the information retrieval endeavour, i.e. the uncertainty introduced by imperfect representation of both documents and queries, and the uncertainty introduced by user-dependence. It should also accomplish all these purposes not only in a single-medium environment, but in a setting where documents may be composed by parts expressed by different media.

This can be summarized, altogether, by saying that the notion of implication embedded in the logic must correspond to what, realistically, relevance of multimedia documents to the request of a specific user amounts to.

Besides being expressively adequate for the tasks, this logic should be computationally tractable, i.e., amenable to being implemented by means of an (either deterministic or probabilistic) algorithm with low computational cost. A second objective of WP4 is then to produce a prototypical implementation of the MIRLOG, that will be the subject of evaluation, both from a theoretical and from an experimental point of view.

WP4 - Approach

The approach that will be followed in the integration of the single contributing theories is to make full use of the potential of denotational, Tarski-style semantics, a standard way of specifying the semantics of representation languages that has been adopted throughout Work Parts 1 to 3. The fact that the algorithms investigated in the single Work Parts contributing to WP4 will all comply with the denotational semantics of the contributing theories, will positively influence the complexity of the integration task.

WP4 - Expected results

The result of Task T41 is the MIRLOG logic and a prototypical implementation of it that will be subject of evaluation within Work Part 5.

Work Part 4 is further structured into Tasks T41 and T42. We now give a concise description of the objectives, approaches taken, and results expected from each of these tasks.

T41 - Objectives

The objective of T41 ``An integrated logic for multimedia information retrieval'', a logic that will encompass in a unifying view the theories developed within Work Parts 1 to 3. The MIRLOG logic should allow the representation and reasoning on the structure and the content of documents and queries, in a way that captures the notion of relevance of documents to users' requests. Besides being expressively adequate for the task of document and query representation, this logic should be computationally tractable, i.e., amenable to being implemented by means of an algorithm with low computational cost.

T41 - Approach

The approach that will be followed in the development of MIRLOG is to make full use of the potential of denotational, Tarski-style semantics, a standard way of specifying the semantics of representation languages that has been adopted throughout Work Parts 1 to 3. This will involve defining the representation language of MIRLOG as the union of the representation languages of the single contributing theories, the ontology of MIRLOG as the union of the ontologies of the single contributing theories, and designing a systematic mapping from the obtained language to the obtaine ontology that complies with the semantics of the contributing theories.

T41 - Expected results

The result of Task T41 is the MIRLOG logic, i.e. a representation language for representing multimedia documents and queries, and endowed with a denotational semantics and an inferential algorithm that complies with this semantics.

T42 - Objectives

The objective of T14 ``Prototyping'' is building a prototypical implementation of inferential algorithms for reasoning in the logic resulting from Tasks T41.

T42 - Approach

An algorithm for drawing inferences in MIRLOG will be designed by combining the techniques adopted for drawing inferences in the single contributing theories. At this stage, it is not possible to foresee what these techniques will be, as they will depend on the results of the investigations done withing the single Work Parts. It is likely, however, that among the contributing techniques an effort will be made to select a unifying one whose inferential power be so comprehensive as to encompass the inferential needs of all the contributing theories.

T42 - Expected results

The result expected from T42 is a prototype of the MIRLOG logic that will be the subject of evaluation (in WP4) against a multimedia document base of realistic size.

The participating (P) consortium members for each of the Tasks in Work Part 4 are listed in the following table.