ReaderBench Talk: Online Conversations Empowered by Advanced NLP Techniques

With the increased need of communication, Computer-Supported Collaborative Learning has gained a broader attention within the last years. Thus, the necessity of using chat applications emerged; however, most solutions lack the integration of Natural Language Processing (NLP) services in order to provide support and guidance, while discussing. Our novel application Talk is integrated within the ReaderBench framework and is defined by three main features: asynchronous messages specific to chats, automated bots for enhancing participants’ involvement and interactions, as well as the usage of advanced NLP functionalities for a posteriori evaluations. ReaderBench Talk communicates with the framework through a dedicated endpoint built within the ReaderBench API. As preliminary validation, two assignments were given to students from our faculty. Overall, students were able to solve the tasks without encountering critical bugs and they also provided valuable feedback. Their feedback revealed the usefulness of the application, its ease of use, the persistency of data, and the quality of the NLP services. The feedback also helped identify issues, which were mainly related to the user interface and to the communication with the server.