Research On Adaptive Presentation Of Personalized Answer Based On Harmony Network

Question answering is one of the important parts in the instruction process. Intelligent question answering depends on the development and application of artificial intelligence and computer technology.This paper analyzed the research status of network-based question answering system based on network home and abroad. According to the requirement of e-learning development, the main goal of the paper is to realize adaptive presentation of personalized answer. Based on learning style, cognitive style and other learning characteristics, a learner model was modeled. On the grounds of the data of learners' characteristics in learner model, personal learning characteristics were obtained by principal component analysis. At the same time, exercises and learners' questions were collected. These were analyzed to find that question style, question kind and answer are related to name and attribute of knowledge point. Then, a question understanding model could be constructed on the basis of knowledge point. Furthermore, a personalized question situation was constructed based on harmony network. The answer parameters, including answer depth and answer presentation pattern, were calculated by harmony function. According to these parameters, the personalized answer was matched by the adaptive neuron-fuzzy inference (ANFI). The system architecture of adaptive presenting personalized answer was proposed in this paper. The system prototype was implemented by using .NET and SQL Server, and tested in practice. The result indicated that the method and technologies presented in this paper are proved effective and reasonable. It makes a good result in question answering of e-learning.