Automated Synthesis of Chatbots for Configuring Software Product Lines

Abstract

Software product lines are a method for creating a family of products that share a typical managed set of features, satisfy the precise needs of a selected domain, and provide an improved quality of software systems by systematically reusing software artefacts at reduced cost and time. A feature model represents the space of all possible and allowed configurations of all products in an SPL. Various predefined feature combinations enable the product to be personalized based on specific user requirements. However, because some features are interdependent and the feature models may have many options, users must understand the implications of selecting the correct feature combinations for the product derivation. Chatbot support can address this challenge by guiding the user through a suitable set of features for the product configuration process. Users can interact with a chatbot using natural language in a familiar environment like Telegram, Slack, or Facebook. In this work, we propose chatbots in the configuration of software product lines based on feature models and present SPLBOT, an approach for SPLs chatbot generators. The methodology relies on Eclipse, FeatureIDE, and CONGA (for Dialogflow chatbot generation). Furthermore, we present an evaluation of our approach’s effectiveness and scalability using three practical examples.