Estimating the importance of consumer purchasing criteria in digital ecosystems

Abstract The purchasing process consists of multiple, non-isolated phases that describe the interactions between consumers and products. One of these phases is the evaluation of alternatives, in which the consumer compares the benefits that will be obtained from each product or service depending on a set of criteria. Recently, the rise of online ecosystems has fueled the development of new purchasing-decision prediction models (PDPM) based on the shared opinions, concerns, and expectations of customers. Commonly, these PDPMs are constructed as multiple criteria decision analysis (MCDA) methods, where experts ascertain the consumer's purchasing criteria, as well as the importance of them, by means of surveys and their own experience. Although relying on experts may be appropriate for identifying criteria, this approach is not suitable for assessing the importance of said criteria due to the large volume of information available on the Web. Furthermore, the preferences of consumers in online ecosystems evolve constantly. In this regard, we consider that the information about purchase-criteria importance should be derived exclusively from the opinions of consumers. However, the attitude and thought-process leading to purchases contain inherent relationships that consumers are not able to express explicitly; consequently, we explore the possibility to make use of the synergies between criteria along user preferences. This paper proposes a new PDPM that estimates the importance of the criteria and values the alternatives by only considering the opinions expressed by consumers in digital ecosystems, while using an expert-defined set of criteria. The proposed approach extracts the weights of both criteria and alternatives and determines the implicit synergies within them without the need of additional input. Lastly, it uses the Choquet integral for weight and valuation aggregation and determines a purchase ranking. This model has been tested in real examples, obtaining satisfactory results.