A diagnosis and monitoring support system for marketing activities

Marketing is an area that aims at satisfying the desires and needs of consumers or groups of consumers (segments marketing), and guaranteeing commercialization of goods and services proposed in optimal conditions of profit. Marketing is thus a set of methods and means at the organization's disposal to promote behaviors, in its target audiences, in favor of the achievement of its own objectives. Marketing therefore aims at forecasting, detecting, causing, altering or stimulating consumers' needs and continuously adapting the production facilities and the commercial structures to the diagnosed needs. We focus our study on marketing activities in e-commerce because the marketing unit of e-commerce has a particularly large source of information at its disposal: hundreds of e-consumers criticisms and evaluations that are continuously gathered and updated through the web facilities. From these criticisms, relevant tendency indicators for marketing analysis could be computed. Indeed, nowadays thousands of people are turning to computational Recommender Systems (RS) before making an e-purchase. RS aim to mediate, support, or automate the everyday process of sharing recommendations between cyberconsumers' communities. They allow them to criticize and evaluate e-retailers according to different criteria such as Range of Services/Products, Product & Service Pricing, Purchase Tracking, Delivery Time, Security of Payment... These so partitioned evaluations databases constitute a gold mine for the marketing unit of an e-commerce. The marketing activity can thus be modelled as a multi criteria analysis of these databases. Furthermore, this analysis must be performed within an uncertain framework due to the dispersion of the opinions of the e-customers. We address these issues with non linear multi criteria aggregation techniques based on the Choquet integral to take interactions between criteria into account and possibility distributions to represent the opinions' dispersion. The global evaluation of an e-commerce is thus perceived as a fuzzy number that results from the aggregation of the partial distributions associated to the evaluation criteria. The marginal contribution of a criterion to the global evaluation of an e-retailer is a basic and useful notion to marketing activities. Indeed, the marketing unit of an e commerce can thus estimate the impact or influence of each dimension of the activity of the e commerce in real time, determine the criteria on which the e-commerce activity should be improved first in order to improve its overall rating as much as possible. Moreover, this principle can be extended to the relative comparison with other competing eretailers. The case of multi modal distributions of e-consumers' evaluations is finally tackled. This principle can easily be transposed in any manufacturing facility to elaborate diagnosis and improvement indexes for a multi-criteria industrial performance synthesized by a Choquet integral aggregation