To address the problem that session-based recommendation models ignore temporal information and spatial relationships among items， leading to an inability to capture complex transition patterns among items accurately， a session-based recommendation model based on time-aware and space-enhanced dual channel Graph Neural Network （GNN） was proposed. Firstly， for the temporal channel， adaptive temporal weights were used to process the items， thereby constructing a time-aware session graph， and the users’ interest-shifting patterns were captured through a time-aware GNN. Secondly， for the spatial channel， spatial relationships among items were embedded into a Graph ATtention network （GAT）， so as to aggregate the information from the perspective of spatial graph structure. Finally， a contrastive learning strategy was introduced to enhance recommendation performance. The results of comparative experiments conducted on three publicly available datasets， Diginetica， Tmall， and Nowplaying — where the proposed model was compared with baseline models including Atten-Mixer （multi-level Attention Mixture network） and GCE-GNN （Global Context Enhanced GNN） — show that the proposed model achieves superior precision （P） and Mean Reciprocal Rank （MRR）. Compared to the suboptimal results， the proposed model has the P@10 improved by 2.09%， 24.97%， and 10.45%， respectively， and the MRR@10 improved by 2.52%， 11.60%， and 4.43%， respectively.

Text classification is a fundamental task in Natural Language Processing （NLP）， aiming to assign text data to predefined categories. The combination of Graph Convolutional neural Network （GCN） and large-scale pre-trained model BERT （Bidirectional Encoder Representations from Transformer） has achieved excellent results in text classification tasks. Undirected information transmission of GCN in large-scale heterogeneous graphs produces information noise， which affects the judgment of the model and reduce the classification ability of the model. To solve this problem， a generative label adversarial model， the Class Adversarial Graph Convolutional Network （CAGCN） model， was proposed to reduce the interference of irrelevant information during classification and improve the classification performance of the model. Firstly， the composition method in TextGCN （Text Graph Convolutional Network） was used to construct the adjacency matrix， which was combined with GCN and BERT models as a Class Generator （CG）. Secondly， the pseudo-label feature training method was used in the model training to construct a clueter. The cluster and the class generator were jointly trained. Finally， experiments were carried out on several widely used datasets. Experimental results show that the classification accuracy of CAGCN model is 1.2， 0.1， 0.5， 1.7 and 0.5 percentage points higher than that of RoBERTaGCN model on the widely used classification datasets 20NG， R8， R52， Ohsumed and MR， respectively.

k-Nearest Neighbor ( kNN) algorithm is commonly used in data imputation. It is of poor efficiency because of the similarity computation between every tow records. To solve the efficiency problem, an improved kNN data imputation algorithm combined with Locality Sensitive Hashing (LSH) named LSH- kNN was proposed. First, all the complete records were indexed in LSH way. Then corresponding LSH ways for nominal, numeric and mixed-type incomplete data were put forward, and LSH values for all the incomplete records were computed in the proposed way to find candidate similar records. Finally, the incomplete records' real distance to candidate similar records were calculated, and the top- k similar records for kNN imputation were found. The experimental results show that the proposed method LSH- kNN has higher efficiency than traditional kNN as well as keeping almost the same accuracy.

BWDSP100 is a 32-bit static scalar Digital Signal Processor (DSP) with Very Long Instruction Word (VLIW) and Single Instruction Multiple Data (SIMD) features, which is designed for high-performance computing. Its Instruction Level Parallelism (ILP) is acquired though clustering and special SIMD instructions. However, the existing compiler framework can not provide support for these SIMD instructions. Since BWDSP100 has much SIMD vectorization resources and there are very high requirements in radar digital signal processing for the program performance, an SIMD optimization which surpported the selection of single or double word mode was put forward based on the traditional Open64 compiler according to the characteristics of BWDSP100 structure, and it can significantly improve the performance of some compute-intensive programs which are widely used in DSP field. The experimental results show that this algorithm can achieve speedup of 5.66 on average compared with before optimization.

Large clusters have been developed to multicore era, and multicore architecture makes new demands on parallel computation. Message Passing Interface (MPI） is the most commonly used parallel programming model, and collective communications is an important part of the MPI standard. Efficient collective communications algorithm plays a vital role in improving the performance of parallel computation. This paper first analyzed the architecture features of KD60 and communication hierarchy characteristics under multicore architecture, and then introduced the implementation of collective communications algorithm in MPICH2 and pointed out its deficiencies. At last, this article took broadcasting as an example, using an improved algorithm based on CMP architecture，which changes the communication mode of the original algorithm. At the same time, this paper optimized the algorithm according to the architecture characteristics of KD60. The experimental results show that the improved algorithm improves the performance of broadcast in MPI.

Since the real node number in cash transport network changes dynamically, a route planning strategy for dynamic cash transport routing was proposed. The strategy did partitioning and optimizing in sequence. Firstly, Dijkstra algorithm was adopted to compute the shortest route between two nodes, and then vehicle number and node on each route were gotten by nearest neighbor algorithm and workload balancing factor. Secondly, the pre-cross genetic algorithm was adopted to optimize each route and get node sequence on the route, which could get the route with shortest distance and minimum time consumption. The experimental results show that the proposed strategy can meet the requirements of dynamic vehicle number and route, and achieve the purpose of saving resources.

The research about concept similarity is very important in knowledge representation and information retrieval. After studying the current classic distance-based semantic similarity algorithm, a more standardized similarity algorithm was proposed by analyzing the other key factors of semantic concept and increasing the impact of the node density and attributes of the concept for the semantic similarity. Through the experimental analysis, the similarity value of the improved algorithm is more reasonable; and compared with human subjective judgements under certain condition of the mediation parameter, the compatibility of the improved algorithm increases about 15% than that of the original algorithm.

With the main purpose of improving the embedding capacity and concealment, using Human Auditory System (HAS) masking properties, an audio steganography algorithm based on lifting wavelet transform and matrix coding to embed secret information was proposed. MPEG Ⅰ audio psychoacoustic model 1 was used to control embedding frames, middle and low frequency coefficients of lifting wavelet transform was choosen to host audio signal, and matrix coding which could improve embedding efficiency and decrease modified proportion was exploited to realize the secret information hiding. The simulation experimental results show that the algorithm not only has excellent concealment and embedding capacity, but also good robustness against noise addition, low pass filtering, resampling, MP3 compression and synchronization attack. Meanwhile, the method can realize blind extraction.

Some problems of the current measures for association rules were analyzed. A new measure named consult was defined and added to the mining algorithm for association rules. According to the value of consult, association rules were classified into positive, negative and invalid association rules. The new algorithm could find out the negative-item-contained rules. Finally, the algorithm was evaluated and analyzed through experiments and practices.

Temporal representation and reasoning is a main research topic in artificial intelligence. Most common models can only represent certain temporal information, but many events happen with uncertain temporal information in real life. A new model for representing uncertain and certain temporal information was proposed to describe events and facts with time indeterminacy. This model firstly defined some temporal objects (such as time point and time period), then defined several relations among temporal objects and discussed the transitivity between the relations. Finally, two examples were analyzed, using this model to solve the uncertain temporal reasoning problem.