In social tagging systems, people can annotate arbitrary tags to online data to categorize and index them. However, the lack of the "a priori" set of words makes it difficult for people to reach consensus about the semantics of tags and how to categorize data. Ontologies based approaches can help reaching such consensus, but they are still facing problems such as inability of model ambiguous and new concepts properly. For tags that are used very few times, since they can only be used in very specific contexts, their semantics are very clear and detailed. Although people have no consensus on these tags, it is still possible to leverage these detailed semantics to model the other tags. In this paper we introduce a random walk and spreading activation like model to represent the semantics of tags using semantics of unpopular tags. By comparing the proposed model to the classic Latent Semantic Analysis approach in a concept clustering task, we show that the proposed model can properly capture the semantics of tags.

Link prediction is a basic problem in the research of social networks. At present, most link prediction algorithms are based on the features extracted from network structure, few research concerns the effect of natural attributes of nodes for creating a link. In this paper we develop a novel way to predict links based on Random Walk algorithm using the information from both the network topology and rich node attributes. The experiment result show that our method can help improves the prediction accuracy and it proves that node attributes have a real effect on link creation.

Nowadays, a great deal of educational data has been produced by E-learning system and MOOC. Educational data is important for Teaching and research. These educational data can be classified many kinds of feature, such as demographic features, social features and behavioral features. And which feature is the most import for student’s performance? In this paper, In this paper, we use some common data mining technologies including Naïve Bayesian(NB), Artificial Neural Network(ANN), Support Vector Machine(SVM) and Decision Tree Classifier(DT) to predict students’ performance, and try to find out the influence of characteristics on students’ academic performance. From the conclusion, we can see that SVM technique outperform others, and Behavioral Features have good effect on students’ performance.

No abstract

No abstract