Evolution Mechanism Of Trending Topics And Its Application In Dual-Structural Network

With the booming of big data, the Internet has become the most important way for people to obtain information. However, ths issue of "information redundancy" and "content governance" is increasingly prominent. The Dual-Structural Network, was proposed in recent years, which consists of a primary structure (the Internet architecture) and a secondary structure (the Broadcast-Storage architecture). By adding broadcast and storing content, the Dual-Structutal network relieves the share bandwidth problem. At the same time, the Dual-Structutal network provides recommendation services to meet the personalized needs of end users. Traditional personalized recommendation algorithms such as content based, collaborative filtering are difficult to help users track their interests. The implicit semantic based recommendation algorithm is more acurrate to characterize users’ interests by mining the end users their interests of topic dimensions. But the topic is constanly changing over time, how to reflect the trending topics’ evolution is of great importance.In order to satisfy personalized recommendation requirements of the Dual-Structural Network, and solve the problem of reflecting evolution of the topics. This dissertation designs the Mechanism of Trending Topic Evolution Tracking, including a new algorithm to track evolution of trending topic called ELDA (Entity based Latent Dirichelt Allocation) and a personalized recommendation algorithm called ETR (Entity and Topic based Recommendation) which based on topics. The main work of the dissertation is reflected in the following aspects:● In order to solve reflecting the topic evolution problem in Dual-Structural Network, this dissertation proposes an entity based algorithm ELDA to measure popular topic’s evolution. By improving tradiational LDA model, this algorithm mines the topics in web dynamically and highlights the feature changes in topics, and links topic at different times by named entities.● With the help of the ELDA algorithm, this dissertation proposes a topic based recommendation algorithm to provide personalized recommendation services in the secondary structure, the ETR algorithm finds the associated topics by named entities, and helps user track the development of the interested topic and provides recommendation services to end users.● To meet the personalized recommendation requirements of the Dual-Structural network, the prototype system of Mechanism of Trending Topic Evolution Tracking implemented. In this prototype system, both the ELDA algorithm and the ETR algorithm have been tested. The results show that the ELDA algorithm can track the evolution of trending topic and the ETR algorithm can support the function of personalized recommendation.