Section: New Results

Causal reasoning and influence diagrams

Participants : Philippe Besnard, Louis Bonneau de Beaufort, Marie-Odile Cordier, Yves Moinard, Karima Sedki.

This work stems on [23] , [24] , [25] , [26] , [27] and, for the logic programming translation, on [53] , [54] . It is related to diagnosis (observed symptoms explained by faults).

The previously existing proposals were ad-hoc or, as in [29] , [41] , they were too close to standard logic in order to make a satisfactory diagnosis. Our proposal starts from a restricted first order logic (of the Datalog kind: no function symbols) and introduces causal formulas, built on causal atoms such as ($\alpha$ causes $\beta$) intended to mean: “$\alpha$ causes $\beta$”. The system is described thanks to these causal formulas, classical formulas, and taxonomy atoms such as ($\alpha$ IS_A $\beta$) ($\alpha$ is of kind $\beta$).

The system produces explanation atoms of the kind ($\alpha$ explains $\beta$ if_possible $\{{\gamma }_1,\cdots ,{\gamma }_n\}$), meaning that $\beta$ can be explained by $\alpha$ if all the ${\gamma }_i$'s are possible together in the context of the given data.

This year, we have improved our logic programming translation in ASP. The aim is to improve efficiency and also reduce the work of the programmer, taking advantage of the declarative aspect of this type of programming. We have applied some of these improvements to two classic riddles, in order to illustrate the power and limitations of current answer set programming systems, and we proposed a few improvements which could make the present systems yet easier to use [12] , [13] .

We are starting a work with some similarities to automatize the treatment of cognitive maps. The aim is to extract relevant information from these maps, which means: building a graph formalism for representing mixed causal and influence relations, and defining a framework (argumentation theory is a good candidate) to aggregate the graphs and provide inference rules in order to infer new information and relations. This work is done in the framework of the RADE2BREST project, involving Agrocampus Ouest and CNRS (GEOMER/LETG), funded by "Ministère de l'Ecologie"(This project is not mentioned in section 8.1 because DREAM is not a partner of this project.). The goal of this project is to model shellfish fishing in order to assess the impact of management pollution scenarios on the Rade de Brest. The cognitive maps result from interviews with fishermen.