The emergence of cloud computing brings a huge impact to software organizations and software architecture design. This new environment implies that the methods, techniques, tools and concepts of the software testing should also change. This paper summarizes the current software testing research issues on cloud computing, discusses the definition, features, classification and research scope of cloud testing, introduces the test platforms and research projects in academia and industry, and analyses the research work of software testing service in the cloud. At last, it points out the contents that should be solved pressingly in software testing on cloud computing.

This paper gives a complete definition of the TaaS, compares the difference between TaaS and traditional software testing model, then proposes the TaaS architecture level model on the basis of international software testing standards, explores the TaaS testing process and its application situations. It sums up the impact factors for user to select TaaS through empirical investigations, and explores the new software testing quality evaluation methods.

In the new generation of software test mode—Test as a Service(TaaS), the definition of test data has changed. This paper proposes a new definition of the test data, explores the reasons for bringing about the change, analyzes the contradiction between security and open requirements of test data through establishing data management model on summary of the characteristics of test data, and gives some advices on test data management policy.

The objective of this study is to evaluate security issues of cloud testing. Organizations are increasingly moving to the cloud and sharing their information, but there are concerns about security issues and risks that may arise due to security breaches. This paper extends the concept of security approach during testing. It makes use of a literature review to evaluate the risk of eight security issues when testing in public clouds. The security issues including trust, governance, compliance, identity and access management, availability, data security, instance isolation and its failures, and architecture in testing on a public cloud. The paper also includes an empirical survey to evaluate the current security strategies of different public cloud providers on the eight security issues.

This paper makes research on a solution to the construction of cloud computing platform for software testing. It first proposes the hierarchical model and the system architecture, and then discusses the approaches to building a cloud computing platform for software testing by using the open source software techniques, including IaaS and PaaS. In implementing these approaches, the authors first list out the software and hardware specifications of the IaaS platform, then describe the installation and configuration of the OpenStack, and illustrate the configuration of virtual testing servers. The authors also present the deployment of the PaaS, which includes the design and implementation of software testing project management tool on cloud computing, the calling method of IaaS resources. At last,the running example of system is given.

The experiment of mining the category knowledge from English-Chinese humanities and social sciences parallel corpus in phrase level is performed based on the established clustering algorithm. The clustering and morphological conversion algorithms are determined by experimental data and specific research needs. The performance of English-Chinese bilingual word features is better than monolingual word by comparing the performance of the Chinese, English and English-Chinese word level knowledge clustering. The category knowledge is directly applied to knowledge base and machine translation system, and the English and Chinese word's expression is explored in mining the category knowledge.

This paper first studies rule-based method and the statistic-based method of named entity,analyzes the strengths and weaknesses of the two methods.Then it presents a cascaded model for named entity recognition, which combines of the rule-based method and the statistic-based method.In the named entity recoginzation experiment for field of defense,the F value is more than 89%.

This paper promotes a method which detecting new technology term from the texts of Chinese patents. Firstly, the element of terms in patents are extracted by ICTCLAS segmentation system and stop words lists. Then the hot elements of terms are filtered based on terms weights computing by improved TFIDF model. Secondly, the hot elements of terms are combined orderly by computing the distance between two words, and obtain the terms collection by terms weights computing and threshold filtering. The valid new technology terms are detected by the experts artificially. Finally, the availability of the method is proved through the applied example.

In order to get similar patents from patent databases, based on merging similar concepts on Ontology-based identification replacement, this paper constructs a vector space model by estimating the TFIDF values of the concepts, and eventually gets a collection of similar clause on similarity matching sentences by orderly longest common subsequence identification method to match the results of previous similar patent document extraction. This method can help enterprises intelligence detect similar patents effectively.

The unfocused problem of network public opinion becomes more and more serious. This article proposes a focused crawler for network public opinion based on content topic selection strategy with time and spatial dimension factor by analyzing feature and evolution of network group events. The results of experiments prove that this focused crawler has higher execution efficiency, and also achives good focused ability. That provides the focused resources of processing network public opinion group events.

In this paper, the authors design a novel citation analysis system based on the full-text literature, which can be used to identify citation location, cited references, and citation context as well as other citation information in the full text. The extracted information is imported into a relational database for the following citation retrieval and citation analysis. A case study is conducted using the field of Nano-tube, cited retrieval is tested firstly and then an analysis based on the search results is conducted. This experiment shows the function and the effect of the system and proves its performance on citation analysis in micro-level and based on full-text data.

With Microblog as the investigated object, based on deeply collecting and analysing the existing literatures, this paper attempts to develop an integrated model that incorporates features of microblog, system quality and subjective norm into the expectation-confirmation model. This research model is examined by questionnaire survey and Structural Equation Modeling(SEM), and the result reveales that users satisfaction, perceived enjoyment, privacy risks and etc. have direct relationship with microblog users' continuance behavior.

There are some shortcomings of hot topic detection in literature,such as single index and the inefficient filtering of high-frequency common words. The paper applies lifecycle theory and TF*PDF algorithm to literature detection, which finds the hot words by tracking the variation of words over time, then locates the time hot words appeared. The results of the empirical tests show that this approach is effective in filtering high frequently used terms and identifying hot research topics in time windows.

The paper proposes a visualization method of correlative relationship between academic research body and content.Correspondence analysis is improved to calculate correlation among research bodies and contents,then network graph is used to show the output,and the research background,method and process are also elaborated.At last,an empirical illustration about information visualization and knowledge map in the intelligence and digital library science is put forward.