In view of the friend recommendation problem in social networks, a friend recommendation algorithm based on the theory of three-degree influence was proposed. The relationships between social network users include not only the mutual friends, but also the other connecting relations with different lengths. By introducing the theory of three-degree influence, the algorithm took all the relationships within three-degree between users into account, while not only considering the number of mutual friends between users as the main basis of the friend recommendation. By assigning corresponding weights to connections with different distances, the strength of friend relationship between users could be calculated, which would be used as the standard for recommendation. The experimental results on Sina microblog and Facebook show that the precision and recall rate of the proposed algorithm are improved by about 5% and 0.8% respectively than that merely based on mutual friends, which indicates the better recommendation performance of the improved recommendation algorithm. It can be helpful for the social platform to improve the recommendation system and enhance the user experience.
To solve the gesture segmentation deviation problem under the interference of other skins and overlapping objects, a method of using depth data and skeleton tracking to segment gesture accurately was proposed. The minimum circumscribed circle, the average and the maximal inscribed circle of convexity defect, were combined to improve the detection of palm and the palm region's radius of various gesture. A fingertip candidate set was got through integrating the finger arc with convex hull, then real fingertips were obtained with three-step filtering. Six gestures have been tested in four transform cases, the recognition rate of flip, parallel, overlapping are all higher than 90% but the rate decreases obviously when tilting more than 70 degree and yawing more than 60 degree. The experimental results show that the accuracy of the proposed method is high in a variety of real scenes.
An evaluation algorithm based on HRank was proposed to evaluate the users' influence in microblog social networking platform. By introducing H parameter which used for judging the scientific research achievements of scientists and considering the user's followers and their microblog forwarding numbers, two new H-index models of followers H-index and microblog-forwarded H-index were given. Both of them could represent the users' static characters and their dynamic activities in microblog, respectively. And then the HRank model was established to make comprehensive assessment on users' influence. Finally, the experiments were conducted on Sina microblog data using the HRank model and the PageRank model, and the results were analyzed by correlation on users' influence rank and compared to the results given by Sina microblog. The results show that user influence does not have strong correlation with the number of fans, and the HRank model outperforms the PageRank model. It indicates that the HRank model can be used to identify users influence effectively.
According to the problem of premature convergence and local optimum in Firefly Algorithm (FA), this paper came up with a kind of multi-group firefly algorithm based on simulated annealing mechanism (MFA_SA), which equally divided firefly populations into many child populations with different parameter. To prevent algorithm fall into local optimum, simulated annealing mechanism was adopted to accept good solutions by the big probability, and keep bad solutions by the small probability. Meanwhile, variable distance weight was led into the process of population optimization to dynamically adjust the "vision" of firefly individual. Experiments were conducted on 5 kinds of benchmark functions between MFA_SA and three comparison algorithms. The experimental results show that, MFA_SA can find the global optimal solutions in 4 testing function, and achieve much better optimal solution, average and variance than other comparison algorithms. which demonstrates the effectiveness of the new algorithm.
In order to improve the real-time response ability of massive data processing, Storm distributed real-time platform was introduced to process data mining, and the Density-Based Spatial Clustering of Application with Noise (DBSCAN) clustering algorithm based on Storm was designed to deal with massive data. The algorithm was divided into three main steps: data collection, clustering analysis and result output. All procedures were realized under the pre-defined component of Storm and submitted to the Storm cluster for execution. Through comparative analysis and performance monitoring, the system shows the advantages of low latency and high throughput capacity. It proves that Storm suits for real-time processing of massive data.