Content of Service Computing in our journal

Published in last 1 year |  In last 2 years |  In last 3 years |  All Please wait a minute... For Selected: Download Citations EndNote Ris BibTeX Toggle Thumbnails Select Research on Service Recommendation Method of Multi-network Hybrid Embed-ding Learning WANG Xuechun, LYU Shengkai, WU Hao, HE Peng, ZENG Cheng Journal of Frontiers of Computer Science and Technology    2022, 16 (7): 1529-1542.   DOI: 10.3778/j.issn.1673-9418.2101032 The network embedding method can map the network nodes to a low-dimensional vector space and ext-ract the feature information of each node effectively. In the field of service recommendation, some studies show that the introduction of network embedding method can effectively alleviate the problem of data sparsity in the recom-mendation process. However, the existing network embedding methods are mostly aimed at a specific structure of the network, and do not cooperate with a variety of relationship networks from the source. Therefore, this paper proposes a service recommendation method based on multi-network hybrid embedding (MNHER), which maps mul-tiple relational networks to the same vector space from vertical and parallel perspectives. Firstly, the social network of users, the shared network of service tags and the user-service heterogeneous information network are constructed. Then, the hybrid embedding method proposed in this paper is used to obtain the embedding vector of users and services in the same vector space. Finally, the service recommendation is made to target users based on the embed-ding vector of users and services. In this paper, the random walk method is further optimized to extract and retain the characteristic information of the original network more effectively. In order to verify the effectiveness of the method proposed in this paper, it is compared with a variety of representative service recommendation methods on three public datasets, and the F-measure values of the service recommendation methods based on single relational network and simply fused multi-relational network are improved by 21% and 15%, respectively. It is proven that the method of multi-network hybrid embedding can effectively coordinate multi-relationship network and improve the quality of service recommendation. Table and Figures | Reference | Related Articles | Metrics Abstract （344） PDF （202） HTML （50） Select Data Set Construction Method for Intelligent Health Care and Its Application ZHANG Linyu, TU Zhiying, HANG Shaoshi, ZHANG Bolin, CHU Dianhui Journal of Frontiers of Computer Science and Technology    2022, 16 (7): 1543-1551.   DOI: 10.3778/j.issn.1673-9418.2101028 The rapid development of Internet and computer technology makes it possible to improve smart health care services in today’s aging population. However, there are some data problems that seriously restrict the process of intelligence in the field of elderly care, such as the lack of real data, the interference of dirty data, and too few standard samples. To solve the problem of lacking data set, this paper proposes a three-stage data set construction method based on machine learning on the basis of small sample data which are collected from the community health care in a city. In the first stage, this paper designs a tree structure-based generation strategy to generate the basic attributes of the data set according to the distribution of the original data. In the second stage, this paper obtains the basic behavioral ability evaluation index of the samples with naive Bayesian algorithm. In the third stage, this paper constructs a variety of multiple linear regression equations to get high-order behavioral ability index and evaluation stage on the basis of the first two stages. In order to verify the effectiveness of the data set generated by the model for downstream tasks, this paper designs multiple rehabilitation training plan recommendation models based on the generated data with neural network, and achieves 5 multi-classification tasks and 2 multi-label classification tasks. This paper verifies the authenticity and validity of generated data through analysis of experimental results and expert knowledge. Table and Figures | Reference | Related Articles | Metrics Abstract （321） PDF （182） HTML （88） Select Structured Prediction Method for Small Sample Workload Sequences LIU Chunhong, ZHANG Zhihua, JIAO Jie, CHENG Bo Journal of Frontiers of Computer Science and Technology    2022, 16 (7): 1552-1560.   DOI: 10.3778/j.issn.1673-9418.2101031 Accurate workload prediction is the key to realize elastic resource management of cloud platform. Aiming at the problem that a large number of tasks with short running time achieve prediction in the cloud platform, which leads to the lack of training data of the forecasting model, a structured prediction of multivariable workload sequences (SP-MWS) method is proposed. It is based on the characteristics of intrinsic correlation among multiple resources consumed in the running of a single task, and the relationship of multi-dimensional workload sequences is explored to supplement the prediction information of small-scale sequence. Firstly, in order to obtain the related workload types, the maximum information coefficient (MIC) and information entropy are adopted to measure the correlation, and related workload types are selected. Secondly, for the selected multiple related workloads, trace-norm regularization multi-task learning (TNR-MTL) is introduced to construct prediction model to realize structural information mining of related workload sequences and complete prediction of multiple workloads simultaneously. Validated on Google cloud platform’s operational monitoring log dataset, the experimental results show that the proposed method can significantly increase model information; the decision-making basis of the prediction model is interpreted and the contribution of each variable to the prediction model is visualized. Comparative experiments show that, the proposed prediction method is better than the commonly used workload prediction methods in time performance and prediction accuracy. Table and Figures | Reference | Related Articles | Metrics Abstract （284） PDF （229） HTML （84） Select Multidimensional Information-Based Web Service Representation Method ZHANG Xiangping, LIU Jianxun, XIAO Qiaoxiang, CAO Buqing Journal of Frontiers of Computer Science and Technology    2022, 16 (7): 1561-1569.   DOI: 10.3778/j.issn.1673-9418.2101024 With the development of service-oriented architecture (SOA) technology, the amount of Web service is increasing. Clustering or classifying Web services correctly are an effective way to improve the quality of Web service discovery and the efficiency of Web service composition. However, the existing Web service modeling methods (such as latent Dirichlet allocation topic model) are difficult to obtain accurate and effective Web service representation from a sparse Web service dataset for Web service clustering. To solve this problem, this paper proposes a multi-dimensional information-based Web service representation method (MISR). First, it generates word vectors which contain topic and semantic information implicit in Web service description with Gaussian mixture model and Word2Vec. Then, the MISR algorithm combines tag-word relationship, popularity, and co-occurrence information together for generating multi-dimensional information Web service representation. Web service clustering and Web service classification are used for evaluating the effectiveness of MISR. Based on a real-world dataset of API services, the experiment results show that compared with LDA, Word2Vec, Doc2Vec, WT-LDA, HDP-SOM, GWSC, the proposed method has 38.8%, 54.5%, 15.3%, 33.3%, 44.7%, 9.7% improvement in Micro-F1 value. Table and Figures | Reference | Related Articles | Metrics Abstract （286） PDF （142） HTML （6）