Is there such a thing as too much data? If not, who is going to be responsible for selecting what we keep? There is only starting to be a profession of data curation. Data curation will need at least three skills: expertise from library, archive and museum studies about choosing, preserving and explaining to users; expertise from computer science and engineering about data processing, data exploration and data storage methods; and expertise from the subject area of the material, so as to know what the data means, where it came from, and what its significance is. Will we do this work with a committee, or train one person to do everything; and if the latter, is that person likely to start from the library, computing, or subject domain?

Currently, the retrieval functionalities of most subject service websites have such problems as literal strings-based searching, poor retrieval effect,lack of search tips and recommendation mechanism. This paper proposes a method to construct contextual navigation of website retrieval system based on Ontology resources label.Modes of information cue navigation and multidimensional classification navigation are designed and implemented through building Ontology and labeling website resource objects, which can effectively guide users’ retrieval interactivity. The evaluation result indicates that the proposed method has significant improvements in itinerary arrangement, retrieval efficiency and user experience compared to traditional website retrieval scheme.

This paper makes a deep and detailed analysis about three main relationships between words in related thesauri, namely equivalent relationship, hierarchical relationship and related relationship. Furthermore, it refines the coarse-grained relationships into 16 kinds of fine-grained concrete relationships and attempts to extract the features of several main concepts and make a detailed description. Finally, based on the above research, a missile Ontology framework is designed, which can provide a research thought and experimental foundation for Ontology construction of aviation field based on thesauri.

This paper proposes a citation knowledge linking mode based on linked data after analyzing the traditional citation knowledge linking, which integrates knowledge sources of different disciplines to build knowledge network in the Semantic Web environment by using technologies of literature knowledge extraction,literature semantization, knowledge linking publishing,knowledge source linking and obtaining. Finally, an experiment which unites Parliament and OpenURL is given to test this mode.

A business intelligence link analysis algorithm based on semantic similarity is designed for the problem of link lost and semantic heterogeneity in the traditional link analysis algorithm. The algorithm exploits anchor chain text and structure synthetically to solve link lost, uses semantic knowledge presented by domain Ontology to solve semantic heterogeneity. The experiment results show that the model and the algorithm achieve a good expected effect and can raise the accuracy and efficiency of business intelligence analysis.

According to the Chinese word syntactic function distribution, the paper constructs syntactic function distribution knowledge in multi-way tree storage structure base based on Tsinghua treebank. The Chinese word category knowledge is mined by using the K-medoids clustering algorithm of Sparse Feature Clustering based on syntactic function distribution knowledge base.

The paper focuses on some fundamental problems in patent text categorization, including the feasibility of using terms for automatic categorization, the research on claim categorization, and the effect of classes with close-related topics on the categorization result. The research is executed on two Naive Bayesian classifiers, kNN, Racchio and SVM classifier, and cross validation is used for testing. The results of the paper are that terms are better than common features under the same settings, that training a classifier with abstracts can improve the claim categorization results, and that classes with close-related topics result in low precision and hierarchical design of classifier is necessary, correspondingly. The paper provides fundamental data for patent text categorization and can be referred by information analysis and other applications using patents.

A topic propagation model of Internet public opinion based on the BA network and SEIR model is proposed in this paper, which fully considers the influence of the media on topic propagation. Two simulation methods are used to analyze the model. The result of the experiment shows that under the influence of the media, the propagation speed of Internet public opinion topic is increased, and the propagation range becomes wider, at the same time, the relaxation time decreases.

This paper analyses the diffused character of public opinion derived for network emergencies,based on the diffused model of public opinion for network emergencies,then researches on public opinion derived character for network, builds derived public opinion monitoring and early warning model through describing the extent of public opinion derived and making the definition of derivative rate.The feasibility of the theoretical analysis is verified by an example, and it can provide reference for the government to realize network management of public opinion.

As propelled by the rapid growth of text data, it is urgent to utilize automated tools to monitor the user relationship, topic trend and the implying values of the platforms. A new modeling framework based on LDA is proposed to evaluate the social networks automatically. The authors first map the text into topic space, eliminating the uncorrelated information based on topic distribution and user feature, then create an evaluation method from social network analysis perspective, mining the structure of the social network from three aspects including user centrality, topic popularity and community activity. Experiments show that promising results are achieved by the new model.

Based on the complex network, this paper studies the net-mediated public sentiment transfer game behavior in the Internet dynamic state. It sets up the network public sentiment transfer evolution game model, emulates the whole process of evolution game, and gets the stable evolution strategy. As a result, the beginning choice of the strategy has an important influence to the net-mediated public sentiment transfer. With the grown up rate of the game partner who has the strategy of “transfer”, the more of the network link numbers, the quicker the balanced state realizes. And transfer process in the public opinion has stagnant and migrated phenomenon, which eventually makes public sentiment towards a recession with population pressure. This paper also gives some advices to promote and control the behavior of net-mediated public sentiment transfer.

This paper conducts a preliminary investigation of open source analysis tools,selects part of them by openness and scalability and gives a brief introduction to their technical features and functions.After comprehensive analysis of the characteristics of the tools, this paper proposes the combined use of the tools for Chinese literature:the SciMat’s data import and pre-processing modules are expanded based on the export data formats of the main domestic literature databases,so that it can import and process the Chinese literature data directly. Integrated with a number of literature analysis algorithms, NWB and Sci2 are used to analyse literature from multi-scope and multi-level.The visualization of the results can increase the readability of the results,which is conducted by GUESS and VOSViewer. Finally, a case study demonstrates the analysis process and verifies the feasibility of the proposes.

This paper takes aspirin side effects as the research topic, and applies one established association rule (IF Drugs and Chemicals categories/adverse effects AND Diseases categories/chemically induced,THEN Drugs cause Diseases) to collect the bibliographic records in PubMed involved the MeSH association rules. Then, it extracts the co-occurrence sentences that include the targeted entities and the semantic verbs between biomedical entities by the natural language processing method. Finally, 30 semantic verbs describing the relations between drug side effects and diseases are extracted. The research result shows that it is feasible to extract semantic relations based on the co-occurrence and natural language processing methods.

Since the storage of UHF-RFID label is limited, the information stored in the label has to be compressed. However, there is not a uniform compression algorithm in China. Focused on the kinds of barcode length and character, the compression algorithm principle and data structure are developed. This method is tested in some customers,and accessible read-write among customers is realized.

Library inquiry machine has its special use requirements and several common problems during normal use. Combining the staff experience of Nanjing University Library and their experiments, this paper brings up a brand new Linux implementation. This paper discusses every steps in the Linux process of the inquiry system. Ultimately, it realizes various functions like the timer switch, streamlined system desktop, district access of Web browser, and remote system update, etc.