Uma arquitetura para extração de unidades de produtos em lojas virtuais brasileiras

Abstract

CONTEXT: E-commerce grows every year since it offers to the e-consumers an easy way to find the best offers and make their purchases online. To enable this task, there are several processes on the search engines used by e-consumers, allowing the comparison of similar products. PROBLEM: This task of comparing similar products becomes a challenge due to the lack of standardization of the registered information in the WEB. In most cases, product atribute information is described freely, unformatted and directly in the product description, making it difficult to establish a standard for comparison. Another important factor is the growth of Brazilian e-commerce and, consequently, the need to simplify the process of e-consumer analysis. GOAL: Due to the problems presented, this study presents an architecture for automated extraction of product units and two prototypes, the first one is a product attribute extraction engine, and the second, one, a WEB prototype for presentation and comparison of product prices by virtual store. PROPOSED METHOD: In order to reach the main goal of this study, we opted for the experimentation-type research method, where the extraction prototype was compared with the developed study related by Horch, Kett and Weisbecker(2015). Also, prototyping was used for the development of the WEB prototype to show the application of the result of the extraction of products attributes, achieved by the extration engine. In order to validate the WEB prototype, the TAM (Technology Acceptance Model) acceptance model was applied. CONCLUSION: Finally, we present an analysis and discussion of the results, besides the acceptance of the prototype, showing that the architecture is feasible and useful for e-commerce.