This paper reviews the progress of research on provenance in the Semantic Web environment, and introduces provenance models including the Open Provenance Model, Provenir Ontology and Provenance Vocabulary, focusing on the description methods and description capabilities of them. Finally, it discusses the difficulties and new challenges when applying these provenance models in the scientific research environment.

This article describes the architecture of digital library private cloud, and detailly introduces the functions and mechanism about the DLCloud infrastructure,Euca2ools middleware,GridSphere subsystem. Then it describes the establishment of digital library templates, and the running methods based on the architecture of the digital library private cloud platform. Finally, the authors summarize cloud platform for digital libraries, and give some suggestions for later research.

This paper introduces a large-scale data processing method based on cloud computing, builds a dynamic, scalable, cost-effective, easy to use and high-performance computing platform on a large of centralized or distributed inexpensive computer cluster, and creates a cloud computing-based framework for large-scale data processing model. It also discusses the methods and applications in this large-scale data processing environment. The computing platform is set up to verify the computing cluster and the feasibility of this model.

Based on concept lattice theory, this article constructs concept lattice of information and its characteristics in virtual community by formal concept analysis. Conceptual clustering and association rule mining assists screening and recommending “analogical anchors” which is higher in accessibility, and the anchoring effect of former anchor information is intervened. From a combination of epistemological and Ontological perspective, try to explore the correction methods of anchoring effect on information screening in virtual community.

Based on the ICTCLAS dictionary segmentation, this paper proposes a method that extracts relevant concept terminology from the Chinese patent texts by maximum matching and frequency statistics, then computes the weights of the items by TF-IDF and gets the final concept terminology. Finally, it analyzes the results with the sample data extraction experiments.

Aiming at the existing problems in the process of extracting user preferences, a new approach that to organize user generated tags by constructing site-level and user-level tag networks on the basis of social network analysis is proposed. Then, topic based tag documents and topic based user networks are formed. A fine-grained user preference model is formed by computing the similarity between them. The experimental results show that the model is scientific.

According to grammar function distribution, a relatively multilevel and systematic grammar knowledge database based on 973 Treebank is constructed from the standpoint of Chinese parser and linguistic knowledge database construction. The grammar knowledge database consists of Chinese syntax knowledge, instance functions of syntax structure and structures’ syntax knowledge. The 51 390 Chinese word’s functions which include 58 kinds of function are gained. The 3 836 instance functions of syntax structure with 58 kinds of function are extracted. The 26 structures’ with 58 kinds of function syntax function are got. There are 900 Chinese sentences rules in the knowledge database. The grammar knowledge database not only offers the syntax knowledge to the Chinese automatic sentence analysis and linguistic researches, but also provides a model for the large-scale linguistic knowledge database construction.

In this paper, a novel algorithm is proposed to mine feature-words frequent itemsets in text database, in order to obtain high-quality expansion terms for query expansion. This algorithm uses the support to measure the frequent itemsets, and only to mine those frequent itemsets containing original query terms and non- query terms synchronously. It can tremendously enhance the mining efficiency. The experimental results demonstrate that the algorithm is more efficient and more feasible than traditional ones.

Focusing on how to capture sudden events timely and accurately, this paper introduces an idea of the Distance between two Segmental Words(DSW), and devises a model for detecting the sudden events in Internet news. This model mainly comprises two parts, as generating the Hot Element of Terms(HET) and detecting new words. Specifically, it uses the improved TF-PDF algorithm for capturing the Element of Terms(ET),which concerns to generate the Hot Element of Terms, and seeks the status quo of distribution among these terms based on the Distance between two Segmental Words, then with the relatively stable combination among these terms to achieve event detection. Experiment shows that the model has a high sensitivity on detecting the sudden events.

Based on the papers indexed by Web of SCI and EI on the theme of search engine’s log mining, this paper constructs a co-author network of this field, and analyses the centrality, small world features, connectivity and other basic characteristics of this network. Furthermore, it also explores the core cooperative research teams, the team members and their research contents.

After summarizing two objective evaluation methods of data waving method and data independent method, this paper gives a new method of indicator difficulty ratio weighting and evaluates the academic journals with standardization score. The results show indicator weighted ratio method has a good differential capacity and is very suitable for academic journal evaluation. The method can’t be used with a few journals and probability weighting method has some shortcomings.

Excessive PDF downloading is likely to cause service disruption of electronic resources. Based on network analysis of PDF downloading behavior, a prototype monitoring system is designed and implemented with open source software. The accuracy of the system is proved by one year of actual operation.

Virtual reference service is playing an active and important role in library services. Based on the open source software A.L.I.C.E., Tsinghua University Library develops the real-time smart chat robot -“Xiaotu” to provide various services, including reference service, booking searching, self-learning etc, and promotes “Xiaotu” into the social networking site, and achieves good results.

To solve the practical problems about electronic resources management in university library, the paper accounts for the demand of the electronic resources service monitoring system. Adopting the SNMP and Java EE technologies, the author completes the architecture design and procedure design of the Library Electronic Resources Service Monitoring System. The system is lightweight, easy to deploy and operate, perfectly satisfies the practical needs of libraries, and is able to monitor the library electronic resource service platform, database, and the unique Z39.50 service of libraries.

The paper gives a solution to solve uniform identity authentication against multiple authentication sources by setting up a LDAP proxy server. This solution evades the problem that commercial system does not support multiple authentication sources and is appropriate for large-scaled organization.