Mechanism For Information Credibility Based On Meta-Synthesis In Dual-Structural Network

Dual-Structural Network was proposed to relieve the "scale-free rise of share bandwidth" problem and the "content governance" problem in recent years, which consists of a primary structure (the Internet architecture) and a secondary structure (the Broadcast-Storage architecture). UCL (Uniform Content Label) is an abstract of web page in the Dual-Structural Network, and the hash field of UCL can solve authentic information credibility problem but it cannot deal with unilateral information credibility problem. The unilateral information spreads explosively with the rapid development of mobile Internet, which is real but unilateral. It usually misleads users into making unreasonable decisions on account of only describing one-sidedness.Web pages, the containers of information, provide Internet information for users. The unilateral information credibility problem can be solved by Hall for Work Shop of Information Aggregation (HWSIA), which gathers all the web pages by topic filed of UCL to do opinion clustering and opinion weighting. HWSIA helps the users understand all aspects of the information and assists reasonable judgment of unilateral information credibility. According to the requirements and features of Dual-Structural Network, this thesis designs the architecture of HWSIA, including OCA (Opinion Clustering Algorithm) and OWA (Opinion Weight Algorithm). The main contents are as follows:? A new opinion clustering algorithm OCA is proposed by investigating features and requirements of Dual-Structural Network. OCA extracts opinion sentences from the web pages based on opinion density and clusters the opinions sentences based on sentence similarity algorithm. Compared to traditional sentiment analysis algorithms, OCA emphasizes the details of opinion class and ensures the diversity of the clustering results.? Taking advantage of UCL, the opinion weight algorithm OWA calculates the weight of each opinion cluster. OWA gets the weights of opinion sentence credibility by combining the credibility of the web page and the support degree of web page to opinion sentence, and then it obtains the weight of each opinion cluster according to opinion sentence credibility weight. The algorithm stresses the fine grain of text and has fast execution speed, which involves the weight of sentence credibility opposite to document credibility.? The prototype system of HWSIA is designed and implemented to meet the practical requirements of the Dual-Structural Network. Extensive experiments are performed on real-world dataset to verify the opinion clustering algorithm and the opinion weight algorithm proposed in this thesis. Experimental results show that the algorithms are feasible and efficient.