With the rapid development of knowledge graph and its wide usage in various vertical domains, the requirements for efficient processing of Resource Description Framework (RDF) data has increasingly become a new topic in the field of modern big data management. RDF is a data model proposed by W3C to describe knowledge graph entities and inter-entity relationships. In order to effectively cope with the storage and query of the large-scale RDF data, many scholars consider managing RDF data in a distributed environment. The key problem faced by the distributed storage of RDF data is data partitioning, and the performance of Simple Protocol and RDF Query Language (SPARQL) queries is largely determined by the results of partitioning. From the perspective of data partitioning, two types:graph structure-based RDF data partitioning methods and semantics-based RDF data partitioning methods, were mainly focused on and described in depth. The former include multi-granularity hierarchical partitioning, template partitioning and clustering partitioning, and are suitable for the wide semantic categories scenes of general domain query, while the latter include hash partitioning, vertical partitioning and pattern partitioning, and are more suitable for the environments of the relatively fixed semantic categories of vertical domain query. In addition, several typical partitioning methods were compared and analyzed to provide enlightenment for the future research on RDF data partitioning methods. Finally, the future research directions of RDF data partitioning methods were summarized.

Hbase can only use key value query, it is not suitable for multidimensional query of mobile objects in road network, which leads to inefficiency in storing index and query. In order to solve this problem, an efficient HBase indexing framework for Road network Moving objects (RM-HBase) was designed and implemented on the basis of HBase storage structure. Firstly, the upper Hmaster and lower HregionServer of the primary HBase index structure were improved to solve the hot distribution problem of distributed cluster data and improve the query efficiency of spatial data. Secondly, the road network moving object index - Road Network tree (RN-tree) was proposed to solve the problem of "dead space" in space division and improve the query efficiency of road sections in the space at the same time. Then, based on the above improvements of HBase index, the query algorithms for spatio-temporal range query, spatial-temporal K Nearest Neighbor (KNN) query and moving object trajectory query were designed respectively. Finally, the Spatial-TEmporal HBase IndeX (STEHIX) framework based on HBase distributed database was selected as the contrast object, the performance of RM-HBase was respectively analyzed from two aspects of the performance of index framework and the efficiency of query algorithm. The experimental results show that, the proposed RM-HBase is superior to the STEHIX framework in both the performance of data equilibrium distribution and the query performance of spatio-temporal query algorithm, and it is helpful to promote the efficiency of spatial-temporal index for the moving object data in mass road network.

Since Markov chain model based software testing does not consider the software structural information and has low ability of path coverage and fault detection, a new software testing model called software hierarchical testing model was proposed based on the combination of statistical testing and Markov chain model based testing. The software hierarchical testing model contains the interaction between software and external environment, and also describes the internal structural information of software. Besides, the algorithm for generating test data set was put forward:firstly, the test sequences conforming to the actual usage of software were generated; then the input data which covered software internal structure was generated for the test sequences. Finally, in the comparison experiments with software testing based on Markov chain, the new model satisfies the software testing sufficiency and improves the test data set's ability of path coverage and fault detection.

To overcome the limitations of traditional face recognition methods for single sample, an improved gradient face algorithm named Orthogonal Gradient Binary Pattern (OGBP), which is robust to variations of illumination, face expression and posture, was proposed. Firstly, the features of the image samples were extracted by orthogonal gradient binary pattern. Then the feature vectors of each direction were concatenated into the general feature vector for face recognition. Finally the Principle Component Analysis (PCA) method was used to reduce dimensions and the nearest neighbor classifier was used for face image classification and recognition. Experimental results on YALE and AR face database indicate that the proposed method is simple, effective and better than the original gradient face algorithm, and also has better performance in face description for single sample.

Concerning the low accuracy of face recognition systems, a face recognition algorithm based on Histograms of Nonsubsampled contourlet Oriented Gradient (HNOG) was proposed. Firstly, a face image was decomposed with Non-Subsampled Contourlet Transform (NSCT) and the coefficients were divided into several blocks. Then histograms of oriented gradient were calculated all over the blocks and used as face features. Finally, multi-channel nearest neighbor classifier was used to classify the faces. The experimental results on YALE , ORL and CAS-PEAL-R1 face databases show that the descriptor HNOG is discriminative, the feature dimension is small and the feature is robust to variations of illumination, face expression and position.

Traditional Collaborative Filter (CF) often suffers from the shortage of historic information. A transfer-Bagging algorithm based on hybrid classifiers was proposed for question recommendation. The main idea was that the recommendation and prediction problem were cast into the framework of transfer learning, then the users' demand for recommend questions were treated as target domain, while similar users who had applicable historic information were employed as auxiliary domain to help training target classifiers. The experimental results on both question recommendation platform and popular open datasets show that the accuracy of the proposed algorithm is 10%-20% higher than CF, and 5%-10% higher than single Bagging algorithm. The method solves cold start-up and sparse data problem in question recommendation field, and can be generalized into production recommendation on E-commerce platform.