With the widespread applications of Massive Open Online Courses （MOOCs） platforms， an effective method is needed for personalized course recommendation. In view of the existing course recommendation methods， which usually use the course learning records to establish the overall static representation for users’ learning interests， while ignoring the dynamic changes of learning interests and users’ short-term learning interests， a Hierarchical and Phased Attention Network （HPAN） was proposed to carry out personalized course recommendation. In the first layer of the network， the attention network was used to obtain the user’s long- and short-term learning interests. In the second layer of the network， the user’s long- and short-term learning interests and short-term interaction sequence were combined to obtain the user’s interest vector through the attention network， then the preference value of the user’s interest vector with each course vector was calculated， and courses were recommended for the user according to the result. Experimental results on public dataset XuetangX show that， compared with the second best SHAN （Sequential Hierarchical Attention Network） model， HPAN model has the Recall@5 increased by 12.7%； compared with FPMC （Factorizing Personalized Markov Chains） model， HPAN model has the MRR@20 increased by 15.6%. HPAN model has better recommendation effect than the comparison models， and can be used for practical personalized course recommendation.

For various applications in today’s service-oriented enterprise computing systems, such as process-oriented service discovering or clustering, it is necessary to measure the distance between two process models. In this paper, we propose a quantitative measure to calculate the distance or similarity between different structured processes. We first introduce a structured workflow model and transform each process into a process structure tree, and then calculate the process distance and its similarity based on the tree edit distance of two structure trees. The proposed distance metric satisfies three distance measure properties, i.e., identity of indiscernible, symmetry and triangle inequality. These properties make the distance metric can be used as a quantitative tool in effective process model management activities. Experiment studies show that the method is feasible. Compared to the adjacency matrix method, the proposed method is more reasonable due to the semantic distance between different structures is considered.

Today, SIMD (Single Instruction Multiple Data) technology has been widely used in Digital Signal Processor (DSP), and most of the existing compilers realize automatic vectorization functions. However，the compiler cannot support SIMD auto-vectorization with the feature of DSP, because of DSP complex instruction set, the specific addressing model, the obstacle of dependence relation to vectorization non-aligned data or other reasons. In order to solve this problem, in this paper, for the automatic vectorization in the Superword Level Parallelism (SLP) based on the Open64 compiler back end, the instruction analysis and redundancy optimization algorithm were improved, so as to transform more efficient vectorized source program. The experimental results show that the proposed method can improve DSP performances and reduce power consumption efficiently.

In numerous classification methods, although Naive Bayesian (NB) classification algorithm is simple and effective, its attribute independence assumption ignores the correlation among attributes. To consider the influence of the attribute independence assumption, a new grouping technology which clusters the conditional attributes was proposed. This technology not only overcomes the deficiency arising from the attribute independence assumption of the traditional NB classification algorithm, but also reflects the different correlation intensity among attributes when the classification is different. Simulation results on a variety of UCI data sets illustrate the efficiency of this method.

A new dynamic distributed algorithm for network routing was presented. The path genetic operators were used in AntNet, and a new pheromone update rule was achieved. Each chromosome was encoded as a series of nodes that in the path ant had found, and was evaluated with a fitness function. The quality of the solution was enhanced through the computation with path crossover and path mutation as well as the population's unceasing evolution. The simulation results show that the improved algorithm has faster speed of the convergence, also the network throughput is effectively improved, and the average time delay is reduced.

Motion parameterization of complex joint model is the foundation of virtual human motion simulation. The classical parameterization methods of 3D rotation, such as rotation matrix, Euler angle, and quaternion, which bear several drawbacks, cannot resolve the motion parameterization of complex joint model. Based on the improvement of exponential map, a novel motion parameterization method called Dual Exponential Map (DEM) which can depict translation and rotation, linear velocity and angular velocity simultaneously was proposed. The motion description as well as parameters computation capabilities of the DEM were introduced. Finally a simulation on maintenance operation training was implemented. The experimental results illustrate that the DEM method bears powerful capability on derivation, solution to differential and ODEs, optimization control and interpolation.

After summarizing the existing traffic prediction methods, a combination method of network traffic prediction based on a variety of forecasting techniques was proposed. According to the theory of decomposition and reconstruction based on multi-scale of wavelet, the network traffic was decomposed into approximation signals and detailed signals of different scales. Then these signals were reconstructed into several low frequency and high frequency time serials by wavelet. These serials were predicted by Linear Minimum Mean Square Error (LMMSE) and Auto Regressive (AR) models respectively according to their different features, and the predicted results of all serials were combined into the final prediction traffic. Simulation results with the real traffic traces show that the method can more accurately predict the future of the network traffic.