A microservice-based architecture for enhancing the user experience in cross-device distributed mashup UIs with multiple forms of interaction

Mobility and continuous connection entail the emergence of heterogeneous devices with multiple forms of interaction. However, it is challenging for developers and corporations to keep up with the devices and provide applications adapted to them. Besides, better user experiences attuned to users’ needs and desires are increasingly in demand. User interfaces play a major role because they must be distributed through different devices and offer a customized experience for each user–device combination. We take advantage of the component-based applications easiness to build customized interfaces that can give optimal solutions to fulfill the requirements for adapting themselves to cross-device applications with multiple forms of interaction. User interaction on mashup interfaces can generate a great deal of data, which can be analyzed for improving the interaction and usefulness of the applications. In our paper, we have created a microservice-based architecture that generates datasets which contain the user behavior for further analysis. Therefore, the user experience and usability in distributed user interfaces may be improved through prediction models generated from the data. Each microservice autonomously fetches its own data and performs consistently so that it can transform datasets optimally by using feature engineering techniques. Thus, data analysis and algorithms can create accurate yet simple prediction models that provide useful knowledge to enhance the user experience. A REST API web service is added to each microservice to facilitate their communication with other microservices and/or third-party clients. The entire microservice architecture, including feature engineering and RESTful API web services for each microservice, offers an infrastructure to handle and process data interaction of cross-devices applications with multiple forms of interaction. This approach has been deployed in a real mashup application where new datasets have been created, processed and validated.