Research On High Performance Processing Algorithms And Propagation Model Of Network Information

Facing with the increasing and massive network information, if we want to mine the useful information and research information propagation rules, the traditional serial algorithm can't meet the requirement of high efficiency. So the research of high performance algorithms of network information processing and information propagation models received much attention.In this paper, we regarded the network information as research object. On one hand,we proposed parallel algorithms of network information processing based on cloud computing model. On the other hand, we researched the propagation rules of network information based on Agent. The main work is as follows.Firstly, we proposed a parallel algorithm of microblog recommendation based on Apriori. According to the mass characteristic of microblog data, we designed a microblog recommendation parallel algorithm based on association rule mining algorithm Apriori using cloud computing model and tested the parallel performance. Experimental results show that microblog recommendation parallel algorithm we proposed has the better speedup performance and the higher operating efficiency.Secondly, we proposed a parallel network information sentiment analysis algorithm. According to the characteristics of network information, the three main steps of sentiment analysis task are improved in parallel and a parallel algorithm of sentiment analysis for network information is designed based on sentiment dictionary, TF-IDF feature weighting and weighted Naive Bayes classification algorithm using MapReduce model. And it is achieved on the platform of Hadoop. The results show that in the condition of large amounts of data, compared to the serial program, parallel system of network information sentiment analysis has the better speedup. It proves that the parallel algorithm is highly efficient for sentiment analysis of massive network information.Finally, the spread process of the epidemic was simulated and a modeling idea of network information propagation based on Agent was proposed. The whole trend of network information propagation was studied according to the differential equation of the SIR model. Meanwhile, the differences of the individual behavior and interaction rules between individuals were considered. It adjusted the parameters of the model using the Agent's autonomy and reactive, which made the model close to the real situation. Thus the characteristics of the propagation network and the information propagation rules can be inferred.