Educational Bots: Learning through Chatbots

We have developed an educational bot using bot framework V4 with QnA, LUIS* and Knowledge mining techniques to recover information about different courses.

The Educational bot is able to obtain the courses that one user is enrolled because he has inferred all user metadata from the student account. We used QnA Maker to build our knowledge base about the courses. In this case, the user is asking for the current government in the Netherlands.

• When the student needs to know more about Mark Rutte, we are using LUIS (Cognitive Services) that is able to detect the “search” intent with the query “Mark Rutte”.
• Previously, we have indexed all the content related to political sciences using Azure Cognitive Search with OCR and computer vision skills. In this example, the bot calls the Azure Search API and order the results by score (TF-IDF) detecting the documents and images where the query “Mark Rutte” appears.
• In the third element of the carrousel appears and image of Mark Rutte. We used computer vision (cognitive services) during the indexing process, being able to obtain the caption of the images and the bounding boxes where the different objects can appear. In this case the persons and objected detected are Mark Rutte and tie.
• The student selects the second document. As we can see in the view of the document the query “Mark Rutte” appears selected in this one.

* Language Understanding (LUIS) is a cloud-based API service that applies customized machine learning intelligence to a user’s natural language conversation or text to predict overall meaning and extract relevant and detailed information.