The Discovery Of Online Social Network Important Users Based On The Information Dissemination

With the coming of Web2.0network, online social network platform keepsdeveloping with a rapid speed, among which Sina micro-blog is at a leader rank. Thenumber of users surges with the expansion of its market size. Online social networkplatform is not for common users only. Increasing enterprises have joined in becauseof its lower cost and higher profit. With no restrictions of time, space, race or culture,a reasonable combination of online social network and commercial activities will notonly offer enterprises with more economic interests but also improve user experienceof the social network. This thesis discussed one micro blog of China Mobile Co. inSina micro-blog platform to dig out the important users in its dissemination sequences.This thesis proposed an effective Weight-Wap sequence analysis algorithm basedon the traditional Wap algorithm. In order to take the impact of users’ comments intoconsideration, this algorithm introduced Node-Weight parameter which allowsenterprises to set it with a rational threshold according to their advertising andmarketing costs. This Weight-Wap algorithm would lead to a moderate account offrequent sequences and important users. While different from other traditional miningalgorithms in Graph theory, it did not hypothesize the optimal dissemination path.Instead, it simulated and generated more than70,000pieces dissemination sequencesdata as closely as possible to the real data by experiment. This data was stored in atree structure which displayed the dissemination direction well. The thesis analyzedthe comments of each user with the semantic analysis algorithm and defined the ratioof positive comments and negative ones as each user’s Node-Weight value. Thus, itguaranteed that all the important users digged out by the Weight-Wap algorithm had avery positive influence on the promotion of products and enterprises. Meanwhile, thethesis introduced an automatic process based on FileNet to help enterprises apply forthe services of data mining and advertising campaign. Proved by the experiment, thisalgorithm had a better performance in practical significance, accuracy, timecomplexity, direction of the dissemination and elimination of negative influence.