This paper identifies the theoretical constructs of information retrieval in information science and computer science,and builds the theory framework of information retrieval from three perspectives that are user, information and technology. It analyzes the research priorities of the two disciplines and cross-domain of information retrieval theory. Finally,it points out the theory status and development trend of information retrieval in information science and computer science,and provides the research direction in information retrieval theory.

After examining some Ontology projects at home and abroad, considering the distribution of Ontology, the authors use a hybrid pattern to construct Ontology with distributed database. On the basis of the readjusted inter-concept relationship, which is one of advance works, ADORES (A Distributed Ontology Retrieval and Editing System) is developed in ASP.NET platform to build the distributed Ontology library. In this paper, the designing ideas and functions of ADORES is introduced firstly, then the key points in ADORES development and their solution methods are emphasized.Finally, some key codes of the difficult points, such as the export of OWL and SKOS standard description format, realization of concept tree and processing of relation reversibility are demonstrated.

This paper proposes uyghur emotional corpus construction specifications and develops uyghur emotional corpus processing system.Combined with the specific characteristics of uyghur emotional words,this paper proposes the full analysis of characteristics of uyghur emotional words in the text,achieves the uyghur emotional words automatic identification method with CRFs model.Experimental results show that the system can effectively process emotional words corpus, and reduce the manual workload.

The CSSCI information services system can only provide single and linear retrieval, and the retrieval effect is not very good. The main reason lies in the unreasonable data organization background. Therefore, the authors put forward the solution which uses knowledge services platform based on Ontology instead of the existing CSSCI information services system, and provide the services in the level of knowledge, so that to improve the retrieval environment. The paper proposes a general model of information retrieval system based on Ontology, designs a practical knowledge services platform named KRSP_CSSCI_Onto, and gives the basic functions and system framework of the platform. An experimental knowledge retrieval system is developed hereby, and the applications of Ontology in knowledge retrieval services is expounded in detail, including semantic extension of search expression, related knowledge recommendation of query results, knowledge navigation based academic resource networks model and knowledge relationships retrieval based rule library.

The paper tries to change the traditional analysis mode of academic resources evaluation, and puts forward the “knowledge-driven” analysis idea which establishes a complete knowledge base firstly, then obtains conclusions on the basis of this knowledge base. For this, the paper adopts the Ontology mechanism to achieve the knowledge organization for CSSCI academic resources, which describes the periodicals and related knowledge by object-oriented approach, so that to establish CSSCI Academic Resource Networks Model (CSSCI_Onto). On this basis of academic knowledge annotated in the Ontology, not only the relationship and association degree between periodicals can be learned comprehensively, but also the bidirectional and multiple association patterns that are implied in original knowledge can be mined, and further the set of core periodicals in the disciplinary during certain period of time can be discovered, which provides factual basis for cooperation and division between periodicals, so that to promote a reasonable distribution of research content of periodicals. The paper focuses on the construction and analysis for relationship of periodicals, which is a part of academic resources association analysis based on CSSCI_Onto.

In order to promote the academic cooperation and development, provide factual basis for the inter-scholar cooperation and obtain analysis conclusion for supporting decision, this paper tries to introduce the Ontology mechanism into the knowledge organization of CSSCI academic resource for organizing concepts related to scholars knowledge map by object-oriented approach, so that to establish scholars knowledge map concept model based on CSSCI. Then the association analysis and knowledge mining are used to discover the potential cooperating probability between scholars and find the related authors who can influence the core authors academically in some discipline field. At the same time the authors group is classified definitely based on the relationship, which can strengthen related field’s research and cooperation, and realize knowledge complement as well as achievement reference and inspiration.

On the basis of summary of “true or false tunnel” strategy on focused crawler, this paper proposes a new KBES algorithm to solute the “false tunnel” problem. The experiments prove that KBES algorithm can improve the efficiency to predict the relevance of new links by anchor and link text in the heuristic strategies to some extent.

For retrieval accuracy problems is due to the diversity and ambiguity of the tags created by users in Folksonomy system,this paper builds a model of Ontology-based Folksonomy semantic retrieval, and analyzes the key problems,such as building Ontology based on Folksonomy and using Ontology-built to implement semantic retrieval.Then current researches are classified and introduced in detail, and finally it makes a prospect for research trends.

This paper firstly creates Medical Subject Headings (MeSH) feature vector for medical literature, then searches for the latest cited times of each paper after published two years, which serves as the indicator for hot papers. Using a genetic program, a group of MeSH terms representing the content of hot papers are extracted. The method is tested on the topic of neoplastic stem cells,and the results from term frequency analysis are served as a baseline. The outputs of two methods are evaluated by domain experts,and the results show that genetic algorithm based method has a better performance than term frequency based method.

Based on deep analysis-of the reasons for the dynamic of the human intelligence network, this paper proposes a new human intelligence network construction method, which integrates the thoughts of knowledge management into it. Using knowledge Ontology to direct the construction of the human intelligence network can improve the soundness of the human intelligence network effectively.

Due to insufficient reasoning abilities, many implicit social relationships existing in the social relationship Ontology are not mined. This article integrates SWRL with Jess reasoning engine. Based on the SWRL rules, the implicit relationship automatic mining experiment is done, and 22 relation rules are designed as well as more than 50 new axioms are reasoned in order to update social relationships Ontology.

The paper firstly analyzes the distribution characteristics of computer education resources on Web, then it designs a multi-layer classifier to resolve the topic classification problem in topic crawling procedure by combining topic words and resources forms, and introduces how to make the precise classification fusion by Naive Bayes Classifier model and how the resources are stored correctly into the hard disk. Finally, experiment results show that the key design idea is feasible and many performances are acceptable, such as the avarage accuracy of the topic classification algorithm reaches to 78% as well as the avarage recall accuracy reaches to 61% and the avarage resources parsing accuracy reaches to 81.5%.

Six basic characteristics, namely man-machine mass collaboration, Semantic Web-based knowledge organization, cross domain knowledge base, multi-language dataset, dynamic database are proposed by literary researches and case studies. Then, the relations between these six characteristics and four typical applications of DBpedia, which are UI, semantic tagging, data mining, cross-domain sharing and servicing applications, are further discussed.

An integrated Mashup-based lightweight application framework is designed and implemented. The framework is easy to expand and loosely coupled integration, which uses a MVC model and the Service Component Architecture (SCA). A digital library portal is constructed based on the framework. And the test shows that the system efficiency is high, and it can quickly integrate all kinds of heterogeneous resources in distributed environment with low-cost. For the middle and small libraries, this solution has a more general applicability.

Based on the background of the second edition Chinese classified thesaurus upgrade project of the National Library of China, this paper introduces the overall situation of bibliographic data update, analyses the rules of bibliographic data update, and provides the program to illustrate each type bibliographic data update.