Principles of AIPioneer Chatbot SystemAIPioneer System is a reasoning machine, which is different from most existing Chatbot systems, Whenever a user starts a conversation, System will create a copy of the Chatbot. Knowledge Database of the copy contains all static Knowledge sets under Ready by the botmaster and all the Rules which status is Ready.
Reasoning machine starts running when the copy is established. Every time a dialog begins, one reasoning will be carried out. Reasoning principle is that System will try to match all the Rule’s conditions in the Knowledge Base. When content in the knowledge Database satisfies a certain condition, system will automatically generate new static Knowledge according to the result of Rule. Therefore, the Knowledge Database of the copy is continually updated and expanded.
User will ask questions or state some facts during the conversation with the chatbot. The system will seek the most appropriate result automatically in the current knowledge database when user ask questions and response with the most suitable answer. When user states, the content of the statement will be added into the Knowledge Database of the copy. The content may also be used for reasoning. The content added by users may also join with the existing content in the knowledge Database to satisfy the Condition of a Rule, and generate new contents. There are several patterns for conversation with the user.
1. User simply ask questions, system replys. This is the simplest pattern. The answer is from the existing knowledge in system. This knowledge may be the static knowledge in the Unit or some new knowledge derived by some Rule according to the static knowledge and the user's previous statement.
2. User states, system memorizes. User inputs a statement, system will add it to the existing static knowledge.
3. Interest trigger. When the user's question or statement Satisfies the Condition of Rule of some kind of interest trigger, system will question the user correspondingly or state this Rule’s some Result.
4. For questions guided by Why and How, system will search for relevant rules in the Rule, and generate answers.