The Navigation System For Intelligent Conversation

Dialogue system is one of the important research direction of natural language processing, in recent years, with the rapid development of speech recognition and natural language understanding, dialogue system is gradually to the practical application, such as apple’s Siri and small I robot, people can express their own thoughts, through natural language and computer intelligent interactive content of a particular field. Its humanized interaction in such aspects as telephone booking, customer service and ticket query has a good application prospect.Intelligent dialogue system of the core and difficulty is the user’s query intention recognition and dialogue management strategy, especially the intention recognition, in a multifunctional dialogue system, directly affects the subsequent processing of the system, and its validity is the most key link.This paper using the template matching and two-stage classifier to analyze user intent. Template matching with regular expressions for pattern matching, the user’s query text conform to return to the intention of the corresponding category; Generalized linear classifier and KNN classifier used for classification.Dialogue management model implements the RavenClaw framework. It isolates the domain-specific aspects of the dialog control logic from domain-independent conversational skills, and in the process facilitates rapid development of mixed-initiative systems operating in complex, task-oriented domains.Our system interactive with user in the manner of question and answer by the techniques like natural language understanding, intention recognition and strategy of dialogue management module. When the query conditions required by the system satisfy the automatic query address database, and will check the address information feedback to the user.