Search

Network embedding transforms a network into a low-dimensional space, in which the information of network can be stored, it is an effective approach to solve the problems of network analysis. However, most of the networks in real scenario are attributed heterogeneous network (AHN), which contains multiple types of nodes and edges. In addition, different types of nodes in AHN have different attribute information. To deal with the heterogeneity of topological structure and node attributes of AHN while preserving the rich information of AHN as much as possible in embedding space, is main difficulty in AHN embedding. To solve this problem, this paper proposes a novel AHN embedding model, namely AHN2Vec (Attributed heterogeneous network to vector), which can preserve node attributes, k-order neighborhood attributes, semantic information and topological structure information in embedding space. Experimental results on two real-word datasets not only show that the overall performance of AHN2Vec is better than the state-of-the-art methods, but also demonstrate AHN2Vec is able to generate high-quality node embeddings for AHN and can achieve excellent performance in downstream tasks.

Tripping quantity of circuit breaker in 10kV rural electric distribution network has characteristic of randomness, and it is inclined to be influenced greatly by various factors such as external environments, power load, power equipment, safety consciousness of power consumer in rural area. A method of prediction based on weighted Markov chain is applied to predicting the tripping quantity of the next mouth in this paper. After transfer of history data series, the rate of tripping quantity of circuit breaker monthly is selected as observed series for predicting. The mean-variance classification model is used to confirm state space of observed series and the self-coefficients of observed series are calculated as weight value by normalization processing. The tripping quantity of circuit breaker in the next month is predicted by the method of weighted Markov chain. Result shows that the method is feasible and has upper applied value in practice.

Having analyzed the drawbacks of image retrieval methods based on content, a novel image retrieval algorithm based on spacial color distribution entropy is proposed in this paper. Firstly, an image is segmented several regions by this algorithm. Secondly, spacial color distribution entropy is designed for each region, which takes not only the local color feature of the region into consideration, but also the spacial color distribution information of the region. With the robustness to rotation and translation, the algorithm avoids shortcoming of losing the location information in color histogram. Through experimental results and comparison with other methods, the proposed algorithm is effective for image retrieval.

Extraction of interesting and general knowledge from large spatial databases is a crucial task in the development of geographical information system. We develop a new method of attribute-oriented spatial generalization. The method is incremental, allowing changes of the spatial database to be reflected in the generalization results without rereading input data. The method also allows fast re-generalization to both higher and lower levels of generality without rereading input. A new data structure, equivalence partition tree, is introduced in the algorithm design of the method. It essentially assures that the algorithm for new method has characters of high efficiency, increment and fast re-generalization. We implemented our new algorithm and ran empirical tests, and we found the method efficient. In addition, its runtime increases only linearly as input size increases.