Research And Design Of Knowledge And Resource Aggregation Based On Collective Wisdom Annotation

Facing large amounts of knowledge and resources,how to mine and characterize the internal relations between knowledge and resources and solve the problem of poor organizational structure and lack of integration of learning resources is a current research hotspot.This study starts with the relationship between knowledge and resources and explores the relationship of knowledge resources through semantic analysis,machine learning and other technologies.The aim is to realize the effective relation between knowledge and resources,and to promote the deep aggregation of knowledge and resources.Firstly,based on semantic analysis technology,the author proposes a hybrid strategy of automatic association between knowledge and resources.It uses a combination of text topic models such as LSI and LDA to mine knowledge topics from rich knowledge entities and transforms knowledge points and learning resources into theme vectors.Through the semantic relationship between vectors,it calculates the similarity between the vector of learning resources and knowledge and realizes the initial correlation between learning resources and knowledge points.At the same time,because of the special relevance of knowledge points and learning resources,and the limited description information of knowledge points,simply using semantic methods to calculate the association weights will have biases in the calculation results.Especially for the case of tacit knowledge,automatic annotation based on semantic analysis is rather difficult.Therefore,this research focuses on the calculation method of correlation weights based on collective wisdom.Using the learner's semantic understanding ability and prior knowledge,labeling the learning resources and knowledge element correlation weights,it realizes the correlation annotation between learning resources and knowledge element by combining the machine learning methods with the the annotation results of crowds.In this study,the related annotation work is assigned to users in the form of tasks.It comprehensively considers user background information and task information and implements optimal matching of tasks and users through two-part correlation graphs.Futher,fusing the results of user annotation,it uses EM algorithm to iteratively estimate the true weight of the task.And evaluates the user confidence based on the results of fusion calculations and user annotation results.Through the above-mentioned mechanism,it can not only guarantee the quality between knowledge element and resource correlation result fusion,but also can truly reflect the user's annotation attitude and ability,and provides a new idea for the relation mining between knowledge element and learning resource.Comprehensive use of automatic annotation and collective wisdom methods to uncover the relationship between knowledge and learning resources and organize learning resources according to knowledge structure.Build knowledge bases and learning resources into an organic,structured whole that enables the deep integration of knowledge and resources.And based on the above mechanism,we designed and implemented a knowledge correlation annotation system,and completed the development and testing of the collective wisdom annotation module and the theme graph knowledge aggregation module.