Software development for decision support

Recommender systems

Several recommender systems in various contexts have been developed based on a multi-criteria approach. A recommendation is made based on the decision maker preferences. These recommender systems are applied to several domains like for example:

  • The determination of an aggregation operator like Choquet, Sugeno, Weighted Sum, Ordered Weighted Sum, Complex Proportional Assessment of alternatives (COPRAS), etc… (see Fomba Soumana Ph.D. thesis defended July 2018);
  • Product configuration. The main idea is here to guide the end-user, i.e. product configurator, in order to minimize the time used to configure a highly complex product having dozen of variables, corresponding to many possible configurations. Our approach is based on the recommendation of values for all variables, taking into account already made decisions during the configuration process. We use a probabilistic graphical model generated from the historical real data on previously configured products. Several experiments were conducted showing that our approach is satisfactory from two points of views: error rate and feedback response time, compatible with an on-line use.

Group Decision

A Group Decision Support System (GDSS) : GRoUp Support (GRUS) based on a multi-criteria approach was developed. The originality of this software lies in the constructive nature of the decision process, based on a multiple criteria approach. Numerous experiments have been conducted using this system in several national and international projects: