A Deep Contextual Product Recommender System for SO-DSPL Framework

The current demand for personalized service-based systems necessitates a thorough understanding of customers’ context and specific needs within the industry. The adoption of Service Oriented Dynamic Software Product Line practices enables companies to craft unique products for each customer. This is achieved by offering a set of interconnected features presented as web services, which are automatically activated or deactivated based on the current operating conditions. These product lines are designed to autonomously adapt to new contexts and evolving requirements. To configure or adapt personalized products, users select the features they desire based on their individual requirements. However, when dealing with extensive feature models, users must comprehend the functionalities of these features and the consequences of their selections in their current context to make informed decisions. Therefore, users require guidance during the product configuration process. To address this challenge, users can express their product requirements using textual language, and a recommended product will be generated based on the described requirements. In this research paper, we introduce a recommendation approach based on deep neural networks that offers personalized recommendations to users, simplifying the configuration and the adaptation process. Specifically, our proposed recommender system is based on a deep neural network that predicts relevant features of the recommended product for the user, taking into account their requirements, non-functional requirements, contextual information, and previous recommended products. Additionally, our proposed framework aims the recommendation of a product adaptation following user context or requirements change. While features are implemented by web/micro services, our approach focuses on service selection by matching the appropriate services to each recommended product’s feature.