Team Orpailleur

Overall Objectives
Scientific Foundations
Application Domains
New Results
Other Grants and Activities

Section: New Results

Around the Kasimir and Taaable research projects

Participants : Fadi Badra, Julien Cojan, Jean Lieber, Thomas Meilender, Amedeo Napoli.

CabamakA and Adaptation Knowledge Acquisition

The research about adaptation within the Kasimir research project is described in  [83] . Adaptation in Kasimir, as well as in many other CBR systems, requires knowledge. The adaptation knowledge acquisition (AKA) is a research work, that can take two directions: AKA from experts (manual) and semi-automatic AKA (using KDD). AKA from experts consists in analyzing adaptations performed by experts. Interviews of experts confronted to decision problems requiring adaptation are analyzed and modeled as adaptation patterns.

Semi-automatic AKA is based on the mining of protocol rules “situation Im1 $\#10230 $ decision ”. Knowing how the decisions change when the situations change from one rule to another rule provides a specific adaptation rule. By generalizing these specific rules, general adaptation rules may be obtained. This generalization process can be implemented through a frequent closed itemset extraction algorithm. The system called CabamakA realizes the mining of protocol rules for adaptation rule acquisition [75] . The KDD process in CabamakA is based on the Coron platform (see § 5.1.1 ). Moreover, an analyst guides the AKA process in CabamakA, using filters to drive the mining process, and interpreting the extracted pieces of information in adaptation rules [12] .

AKA from experts and semi-automatic AKA are not completely satisfying: The “AKA from experts” provides generic adaptation patterns that are intelligible, but cannot be directly operational. The “semi-automatic AKA” provides adaptation rules that can be directly implemented, but that are difficult to understand (and thus, to validate). A possible research work would be to combine these two kinds of AKA for producing operational and intelligible adaptation knowledge units.

New Directions in the Taaable Project

The Taaable project has been originally created as a challenger of the Computer Cooking Contest (CCC, ). A candidate to this contest is a system whose goal is to solve cooking problems on the basis of a recipe book (common to all candidates), where each recipe is a shallow XML document with an important plain text part. The size of the recipe book (about 800 in 2008 and about 1500 in 2009) prevents from a manual indexing of recipes: this indexing is performed using semi-automatic techniques.

The first version of the Taaable system (2008) was the European vice champion of the contest. It has been presented as a demo in [51] . The second version (2009) was the World vice champion of the contest. A third version for the 2010's contest is under conception.

The partners of the 2009's Taaable project are members of Orpailleur, of the SCORE Team at LORIA, and of the SILEX team of the LIRIS (Lyon). Beyond its participation to the CCCs, the Taaable project aims at federating various research themes: case-based reasoning, information retrieval, knowledge acquisition and extraction, knowledge representation, ontology engineering, semantic wikis, text-mining, etc.

A general description of the 2009's Taaable system, also called WikiTaaable, can be found in [35] . The most important original features of this version are:

The current case-based reasoning inference engine of Taaable and what should be its future are described in [56] . Another ongoing work is about the application of minimal change theory to adaptation in case-based reasoning and its future application to Taaable [59] . (major [4] ). The main contribution in this area in 2009 is the application of this approach to a formalism including quantitative values and how it can be reduced, under some assumptions, to linear programming. Finally, a research about temporal reasoning for case-based reasoning with an application to the adaptation of recipes is under study [55] .


Logo Inria