[Objective] Understand the attitudes of Chinese academic journals toward Institutional Repositories deposit policies. [Methods]Select 326 samples from the list of "Report of Chinese S&T Journals Citations", telephone them to make sure that they answer this online survey. [Results]195 valid surveys are returned and the effective rate is more than 65%. 74.9% journals agree that the papers can be stored in the Institutional Repository, so their attitudes and recommendations about the article deposit and communication by Institutional Repositories are positive. [Limitations]The survey scale is limited on the high quality, well-known and digital journals, and the result is not applicable to the whole Chinese journals. [Conclusions] Most of investigated targets agree the policy of article deposit and communication of Institutional Repositories, but as the recommendation of this survey, they wish that the consulting service about the rights of journal article open access is provided in future.
[Objective] The paper uses linked data technology for e-commerce credit information management, which makes users access to the required credit information services quickly, effectively and intuitively. [Methods]Based on establishing the relationship of e-commerce credit information entities and using Protégé to construct credit information Ontology, this paper frames an e-commerce credit information service model and uses AllegroGraph which is a graph database for storage and query. [Results] Constructing an e-commerce credit information service model based on linked data, which realizes a series of management services, such as credit information storing, browsing, querying and result representation. [Conclusions]The service model with strong theoretical and practical value can realize the linking and sharing among the e-commerce credit information data sources.
[Objective] The authors analyze and summarize the characteristics of semantic annotation tools, construct the automatic semantic annotation model framework for digital library resources, to provide references for the design and application of Chinese semantic annotation tools. [Methods] Conclude and analyze the technology and methods of semantic annotation platform framework, combining with the comparative analysis of foreign mature tools, and introduce the system development and the idea of modularization to construct the model. [Results] The automatic semantic annotation system model includes system input module, semantic processing module, Ontology knowledge module, semantic tagging module and storage module. The overall algorithm thought and automatic labeling algorithm of Ontology thought are also analyzed. [Limitations] Relevant algorithms involved in the automatic semantic annotation model are still in development, this article only analyzes the main algorithm thoughts and instructions. [Conclusions] The automatic semantic annotation model can provide references for the design of Chinese semantic annotation tool.
[Objective] To solve the scalability problem and data sparsity problem of the collaborative filtering. [Methods]This paper proposes an algorithm of collaborative filtering personalized recommendation through directionally mining users' preferences. Introducing time as a variable, the algorithm excavates in two stages. The first stage is to find the project-based weak similar users, the second stage is to use users' relevance and attribute similarity so as to do directional excavation and form a collection of recommendation. [Results]Experimental results show that the time complexity of the new algorithm reduces a magnitude. Furthermore, the more sparser the data is, the greater leading advantage the recommendation accuracy has. [Limitations] The algorithm recommends deeply by analyzing the users' existed preferences, and it doesn't involve the users' preferences which haven't appeared. [Conclusions]This algorithm has a strong ability to adapt to data sparsity and enhances its scalability at the same time.
[Objective] To summarize the researches on the information quality evaluation of online Community Question Answering (CQA). [Coverage] Use Web of Knowledge and CNKI and reference retroactive method to retrieve English and Chinese literatures related to information quality evaluation of online CQA. [Methods] Literature investigation, summarized by research topic. [Results] So far, the researches on the information quality evaluation of online CQA sites focus on three aspects that are influencing factors of evaluation, manual evaluation and automatic evaluation, and for automatic evaluation, the machine learning application is the main method. [Limitations] Comparison of different solutions lacks specific quantitative evaluation. [Conclusions] There are some weakness in the current researches such as lacking authoritative evaluation criterion and lacking domain focus. In the future, more comprehensive and deeper researches are needed in this area, and automatic evaluation will be the hot spot.
[Objective] A new service cloud which supports on-demand self-service is built for monitoring strategic S&T information. [Context] Based on the existing automatic Web information monitoring system, the system need extend to support more information analysts. [Methods]With regard to the problems of scalability and flexibility of the existing system, an idea of building a new service cloud is proposed, the service cloud with focusing on six aspects of the problems is designed and implemented. [Results] The service cloud for strategic S&T information monitoring with the characters of scalability and flexibility is implemented and now is used by more users. [Conclusions]Implementation of the service cloud results in an on-demand self-service model for user. The new platform supports more information analysts and provides more effective service for information analysts.
[Objective] This paper generalizes the framework of patent technology mining based on text, extracts the key techniques and analyzes some typical application scenarios. [Coverage]Chooses 105 papers from Elsevier, Springer, CNKI databases and Global TechMining Conference, and refers 66 papers at last. [Methods]Review semantic knowledge representation of patents, analyze the research progress of three typical technology mining scenarios and summarize the hot research topics of patent technology mining based on text. [Results] The result shows that the semantic knowledge representation of patents is very important to patent technology mining. And patent technology mining oriented to problems and solutions based on SAO units will be the hot research topics. [Limitations]Only focuse on the applications in patent technology mining of the techniques (e. g. Text Mining, Statistics and Natural Language Processing), but the development trendency of these techniques need to pay more attention. [Conclusions] This paper will facilitate to give an overview of patent technology mining, the key problems and the typical application scenarios.
[Objective] Taking netizen crowd as the central of the research, the paper aims to discover the inner rule of the public opinion evolution in the unexpected event. [Methods] The study proposes the theory of crowd simulation which takes the individual as the agent to explain the behavior rule of crowd in a certain context and the inner mechanism of evolution of Web public opinion in unexpected events. [Results] Experimental results show that opinion leaders can reduce the evolution hours. And the government has a great impact on putting down the Web public opinion. [Limitations] The crowd simulation of Web public opinion in the unexpected event is fairly simple. Many more factors have not been taken into consideration. [Conclusions] This achievement can be used to forecast the evolution path of Web public opinion in the emergency management of unexpected event, and support the establishment of Web public opinion strategy and information disclosure strategy.
[Objective] To discuss the effect of features quality on the search results through the four major domestic microblogging. [Methods] The weight calculation indicators TF-IDF is enhanced from the perspective of the whole feature, and the quality of each feature in the microblogging is further assessed by the two measure indicators including descriptive power and discriminative power. [Results] The descriptive power and discriminative power in microblogging appeare positive effects on the search results; Different quality of features in each platform has different impact to the classified results; And integrating the two indexes has the best effective in the classification. [Limitations] Some other features in the microblogging, namely dialogue replies, and number of fans, have not been taken into account. And the word semantic ambiguity characteristic like synonyms is not discussed yet. [Conclusions] This study helps features in the microblogging to be in-depth understood through the discussion that the effect of features quality on the search results. So as to improve the retrieval quality in the social media platforms.
[Objective] In order to keep the fidelity of digital information resources in libraries, museums and archives during the process of copyright protection.[Methods] The zero-watermarking scheme based on block compressive sensing and Arnold transformation is proposed. The proposed scheme uses important features of original image and the meaningful copyright information to construct zero watermark. [Results] The acquired zero watermark has good discrimination. Also, this scheme possesses the robust ability to resist on image-manipulation attacks under consideration. [Limitations] Since the subjects in this study are gray level images, there are some limitations. [Conclusions] The optimized scheme using block compressive sensing, in a manner, contributes to the copyright protection of digital information resources in libraries, museums and archives.
[Objective]Base on the principle of diversity and correlation, stimulate the group creativity by recommending pictures. [Context]The application context is group idea generation environment. [Methods] Base on text segmentation, webpage analysis technique and cosine similarity algorithm, the paper establishes a model of picture recommending system and raises Max Difference Algorithm (MDA) to select pictures for groups which in the process of creativity. [Results] This research proves the positive effect of picture recommendation system to group idea generation by an experiment method and proves the usability and accessibility of the system by survey. [Limitations]The Max Difference Algorithm is based on the text information around the pictures, and this information is often in difference with the pictures' content, so there are limitations on the MDA.[Conclusions] In the group idea generation process, group creativity performance can be improved by recommending pictures.
[Objective] This paper aims to address problems of automatic operation and maintenance in university library's data center where a large number of servers work. [Context]The data center of university library is dealing with more and more tasks, which leads to the increasing number of servers. With both the virtualization platforms and traditional servers in the data center, manual work can no longer manage them all. [Methods]Search for suitable open source software; design, deploy, and test the best practices; study on the data structure of open source software for secondary development, and integrate multiple open source software. [Results]Successfully solve the problems which include: location management of servers and virtual machines, network interconnection structure, automatic deployment, IP allocation, monitoring, backup, centralized log management etc., and ultimately manage to demonstrate the readers a current server status page. [Conclusions]With the application of open source software, the library's data center reaches the goal of clearness, standardization, automation and external transparency.