Research On Personalized Just-in-time Learning Ways Based On Constructivist Learning Theory

Personalized recommendation is becoming the basic form of information network services in the Internet+ era.Although its wide use in e-commerce and social media has produced huge commercial value,there are only limited research and applications in the field of personalized knowledge learning,which may have tremendous potential social value.The current personalized recommendation techniques are mostly limited,while still exist key and difficult issues,such as cold start,data sparse etc.In addition,the knowledge recommendation is specific,need to consider the user's own learning ability,and different knowledge recommendation sequence on the user's impact in learning difficulty and learning efficiency.In addition,the knowledge itself is a dynamic evolution of the structural system,which requires that knowledge recommendation service can reflect the evolution relationship between the knowledge,and to provide users with time-evolved knowledge services.Accordingly,this paper aims at the requirements on continuity and evolvability of recommended knowledge sequences in personalized knowledge learning.Firstly,this thesis introduces a kind of personalized knowledge recommendation method based on constructivist learning theory.New model uses knowledge network to model the knowledge system,uses the nearest neighbor priority strategy to select knowledge candidates,and introduces top-K unstudied knowledge recommendation algorithm sorted by descending learnable constructive degrees of knowledge candidates to output recommendation items.In the experimental part,the knowledge of digital media domain is used as the research domain,and the recommended examples of different algorithms are compared and analyzed according to the two evaluation indexes of learning efficiency and knowledge sequence relevance.The experimental results show that the more strong knowledge relevance and knowledge system coverage efficiency can be obtained by proposed model.Secondly,taking into account the knowledge learning is a process of accumulation over time,the feature of knowledge evolution should be worthly concerned with in the knowledge recommendation process.Therefore,this thesis proposes a domain knowledge evolution analysis method based on the joint modeling of space-time domain.Wherein,the new method firstly considers the knowledge system can be expressed in the form of knowledge network with joint space-time correlation,then adopts skeleton clustering method to extract the evolutionary path of the knowledge network.The experimental analysis on the domain of digital media shows that the proposed method can effectively extract the evolutionary path of domain knowledge.The new method can help users to sort out complex knowledge relations,clarify the development process of certain domain knowledge,and has clear significance for knowledge learning and personalized recommendation.