The major components include the following:
- Intent detection
- Conversational design with routines
- Dialogue design
- Bootstrapping and sustaining students’ opinions
- Course information management
IBM Watson Conversation is an online NLP tool provided by IBM. With some pre-defined intents and entities, different rules can be formed. When EASElective receive a natural language inputted by the user, it will be sent to IBM Watson Conversation with the help of the provided API in JSON format. Then, IBM Watson Conversation will match the natural language sent by the EASElective with the defined rules and response.
Techniques and Technologies used
- Natural Language Processing (NLP) Tool
IBM Watson Conversation is a simple chatbot developing tool provided by IBM. Since the structure and the design are too simple with limited database access and routine control, the main control will be a middleware developed by ourselves using Node.js. IBM Watson Conversation is used only for extracting the intent of the user input.
Telegram is an instant message application. It also provides API for user to develop a chatbot on Telegram. Therefore, telegram is used as the user interface.
MySQL is a relational database system. MySQL is used to store the content, course comment, teaching staffs of the course and comment of the teaching staffs, etc. All of these data have relations to each other. So, relational database system is used.
MongoDB is a document-oriented database system. Since the user model will be represented by JSON object, MongoDB is used to store the user which has not talk to the chatbot more than a day.