情 报 学 报

ISSN 1000-0135

第30卷 第7期 675-681，2011年7月

JOURNAL OF THE CHINA SOCIETY FOR SCIENTIFIC AND TECHNICAL INFORMATION

ISSN 1000-0135

Vol.30 No.7， 675-681 July 2011

1.Study on Potential Knowledge Discovery from Electronic Medical Records Based on Semantic Triple

Wen Youkui1,2 and Jiao Yuying1

(1.School of Information Management of Wuhan University, Wuhan 430072；

2.School of Economy and Management, Xidian University, Xi’an 710071)

Abstract: Electronic medical records (EMR) is a challenging task of medical informatics (MI) in order to solve the expression, organization, application in the field of medical information. Information of medical records has specific semantic requirements, mainly in professional represents of medical information and truth of the actual medical procedure. Electronic medical records can not now express dynamic semantics relationship between the concepts of pathological; it is difficult to realize semantic based information retrieval and potential pathological findings. For this reason, we developed a electronic medical records of potential knowledge discovery systems based on knowledge element, which extract electronic medical records into the knowledge element of semantic triples form, to solve semantic reasoning and potential disease problems of the electronic medical records, to implement integrated system of treatment cases for diagnosis, instrumentation detection, treatment options. Experiment proved the system have some guidance for diseases treatment, teaching and research.

Keywords: electronic medical records, knowledge discovery, knowledge element, semantic triple

2. Research on Multi-perspectives Topic Constructing

Wu Qingqiang1，2 and Zhang Xiaolin2

（1.Institute of Scientific and Technical Information of China, Beijing 100038；

3.Library of Chinese Academy of Sciences, Beijing 100094）

Abstract: In this article, the authors conduct the research on the multi-perspectives access to topic structure. Firstly, an experiment is used to discover that there are some obvious differences between the two topic structures generated by the variant analysis and the co-word analysis, and both of which are a reflection of the existing topic structure. According to the characteristics of variant analysis and co-word analysis, this dissertation adopts the relationship integration, the process integration and the result integration to obtain the topic structure in multi-perspectives. In the relationship integration, three different functions including the linear function, the Max function and the co-variant are adopted. The tests have been carried out in terms of the topic class levels, the relationship of words and the relationship in the topic class to verify their adaptation to the topic structure in multi-perspectives.

Keywords: topic constructure, multi-perspectives topic constructing, co-word analysis, variant analysis

3. Research on Semantic Reasoning with Ontology and Rules

Tang Xiaobo and Jin Zhongming

(Center for Studies of Information Resources of Wuhan University, Wuhan 430072)

Abstract: To solve the information share and integrating for related concept of ontology, find the semantic relationship among ontologies, this paper proposes the semantic reasoning model ORRM with ontology and rules combined, establish the family ontology FO. Semantic reasoning focus on two layers, the first layer uses Racer to reason based on description logic, testing ontology conflict. The second layer uses concepts and attributes presented by ontology for member rule base, using SWRL and Jess as inference engine. The new ontology exported by the model adds the semantic relationship between the concepts, makes implicit knowledge explicit and perfects the contents of ontology knowledge base. In the fields of semantic web, the utilization of ontology knowledge is improved by the application of the model.

Keywords: domain ontology， description logic， SWRL， Jess， semantic reasoning

4. Language Model and Its Application to Information Retrieval

Su Sui, Lin Yuan and Lin Hongfei

(Laboratory of Information Retrieval, School of Computer, Dalian University of Technology, Dalian 116024)

Abstract: Language modeling approach to information retrieval is promising and challenging because of its favorable foundations in statistical theory, which can deduce other classical retrieval models easily. However the recent researches on language modeling approach focus on the task of single language retrieval, and there are few studies examining its effectiveness on cross language retrieval. In this paper, we apply language modeling approach to the task of cross language retrieval based on introducing the retrieval approaches of language models. This paper presents two cross language retrieval model: statistical translation model and cross language relevance language model, and also analyzes their effectiveness.

Keywords: statistical language model， cross language information retrieval， relevance model

5. A Collaborative Filtering Recommendation Algorithm Based on Item-Class Preference

Leng Yajun, Liang Changyong, Zhang Enqiao and Qi Xiaowen

(School of Management, Hefei University of Technology, Hefei 230009)

Abstract: Currently collaborative filtering is the most successful and widely used recommendation technology in recommender systems. However, with the development of E-commerce, the magnitudes of users and commodities grow rapidly, which results in the extreme sparsity of user rating data. The method of searching for nearest neighbors in traditional collaborative filtering algorithm works poor in this situation, which makes the quality of the recommender systems decrease dramatically. To address this issue, a collaborative filtering recommendation algorithm based on item-class preference is proposed. The proposed algorithm first finds out a set of candidate neighbors who are similar to the active user in item-class preference. The candidate neighbors have similar interest and more co-rated items with the active user. Then the algorithm identifies some nearest neighbors in the candidate neighbor set, which eliminates the interference of the users who have few co-rated items with the active user, and enhances the accuracy of searching for nearest neighbors. The experimental results show that the proposed algorithm can efficiently improve recommendation quality.

Keywords: recommender system, collaborative filtering, item-class preference, similarity

6. A Study on the Method of Mobile Content Recommendation Based on Frequent Marked Lattice

Cai Shuqin1, Zhang Yu2, Hu Muhai1 and Xiao Quan3

（1.School of Management, Huazhong University of Science and Technology, Wuhan, Hubei, 430074；

2.School of Management,  Wuhan Textile University, Wuhan 430073；

3.School of Information Technology, Jiangxi University of Finance & Economics, Nanchang 330013）

Abstract: In existing mobile content service systems, There are few studies is still quite rare on automatic situation service rule construction. Hence, a method is proposed that the semantic association rules between situations and preferences are built by quantitative frequent marked lattice based on ontological model. The method provides a solution to the problems of rule collision and context data availability. Frequent Marked Lattice further reduces needed node number for rule generation compared to frequent concept lattice and it is more convenient to extract different rules and calculate related parameters. The construction algorithm of FML and priority mechanism of rule extraction are designed. The validity of the algorithm is verified by experiment and an analysis is performed compared with the related works.

Keywords: mobile content recommendation， situation， context， ontology， frequent marked lattice

7. Research on Knowledge Transfer Behavior of individuals Based on Cellular Automata

Wu Jiangning, Liu Na and Xuan Zhaoguo

 (Institute of Systems Engineering, Dalian University of Technology, Dalian 116024)

Abstract: Knowledge exchange and transfer are key means for the organization to acquire knowledge and hence promote its knowledge stock and competitive competence, which has attracted more attention from many knowledge managers. This paper starts with this issue and focuses on the different learning styles of individuals within the organization. The individuals involved are divided into two types, i.e. active learner or passive learner, whose learning attitude, behavior and knowledge stock will be changed during the process of knowledge exchange. The knowledge to be exchanged is expressed by a multi-dimensional vector. Hereby, a cellular automata knowledge transfer model is proposed to study the influence of structure of neighborhood, the ratio of initial active learners, the people movement and expert introduction mechanisms on the performance of knowledge transfer processes by using two kinds of rules which are the state update rule and the knowledge exchange rule respectively. Simulation results show that the local interactions between individuals at the micro-level reveal some complex properties like self-organization at the macro-level within the organization. The structure of Moore neighborhood is more effective on the knowledge exchange and transfer within the organization, and the growth of the average knowledge stock is going up with the increase of the ratio of initial active learners. Both the people movement and the expert introduction mechanisms can improve the performance of knowledge transfer and knowledge stocks. Besides, the growth of the average knowledge stock is proportional to the self-learning ability of individuals.

Keywords: knowledge transfer, cellular automata, people movement, expert introduction

8. Evaluating Influence of Sampling Methods Upon Social Network Analysis

Tan Xiaojie, Wu Kewen, Zhao Yuxiang, Zhu Qinghua and Wang Qianjun

(Department of Information Management, Nanjing University, Nanjing 210093)

Abstract: Social Network Analysis(SNA), a quantitative research method which aims at analyzing the structure and various properties of target network, has been applied in many academic fields. However, those existing studies emphasize more on the explanations of SNA indexes, rather than evaluating the impact of process elements to the research conclusion. This study selected four classic sampling methods (node sampling, link sampling, depth first sampling and snowball sampling) as the research object, and briefly analyzed the impact of different sampling methods to five typical SNA indexes (diameter, average distance, clustering coefficient, degree centrality and betweenness centrality) in scale free networks, which can be helpful to formalize the research process of implementing SNA.

Keywords: social network analysis, sampling method, sample error, snowball sampling

9. A Study on the Scientific Research Collaboration Network of “985 Project” Universities in China

Qiu Junping and Wen Fangfang

(The Research Center of Chinese Science Evaluation, Hubei Wuhan 430072)

Abstract: Universities have become the main component in the national scientific innovation system. As the collaboration between different universities strengthens, the study on the scientific research collaboration is of increasingly vital value and importance. This paper chooses the “985 Project” universities as the sample to study on their scientific research collaboration relationship from the perspective of co-authorship based on the methods and tools of Social Network Analysis. The research results illustrate that the 39 universities has already established the primary scientific research collaboration, but the intensity still waits for enhancement. Besides, there exists significant correlation between scientific output and scientific collaboration. It means that to some extent to strengthen collaboration intensity can improve the quantity of universities’ scientific output.

Keywords: scientific collaboration, collaboration relationship, social network analysis, co-authorship

10. Patent Mapping of Technology Competition among Fortune 500 Enterprises Based on Global Patent Co-citation Analysis

Wang Xianwen1，2， Ding Kun1，2 and  Zhang Xi1

(1.Faculty of Humanities and Social Sciences, Dalian University of Technology, Dalian 116085；

2.Joint Institute for the Study of Knowledge Visualization and Scientific Discovery， DUT?Drexel）

Abstract: Patent citations are generally used to provide support for specific statements of  technology competition, when patentometrics has become one kind of main method to analyze competitive technical intelligence for companies. In this paper, the authors choose Fortune 500 enterprises, released in 2009, as research objects. Using patent data (2000~2009) from Derwent Innovation Index, employing full network methods in social network analysis, the authors construct the global patent co-citation matrix of Fortune 500 companies based on the patent forward citation, and conduct patent mapping analysis using information visualization technology and social network analysis method, including clustering distribution, Kernel density, and co-citation network analysis, to study the technology clusters, technology competition structure of Fortune 500, and to find the pivotal enterprises in the technology competition network.

Keywords: forward citation， patent co-citation， technology competition， patent mapping， global co-citation

11. Co-word Analysis on the Hotspots in Chinese Library and Information Science for the Last Decade

Wang Hong

(Hubei University of Automotive Technology, Shiyan 442002)

Abstract: Five years as a period，this paper uses the co-word analysis method to perform cluster analysis and cluster relationship analysis to the high frequency keywords of eight core journals of Chinese Library and Information Science（abbreviation LIS） from 1998 to 2007. At the same time, the paper draws the relation table of clustering results and drafts the relationship figure of cluster, and systematically analyzes the research hotspots of Chinese LIS in different period, in order to reveal microscopically the research hotspots of Chinese LIS in different period. To sum up, the research hotspots of Chinese LIS has a certain characteristics of inheritance, continuity, stability, expansibility and variability.

Keywords: co-word analysis, library and information science, cluster analysis, cluster relationship analysis

12. The Study on Information Visualization Technology in Medical Information Analysis

Xiong Jun and Qiu Xiao

（School of Medicine and Health Management, Hangzhou Normal University, Hangzhou 310036）

Abstract: We made the scientific map basis on the citation analysis, co-citation analysis, cluster analysis and social network analysis, and with the help of Citespace, Bibexcel, Pajek and Ucinet . Discussion on the use Information visualization techniques in medical information analysis. Through the knowledge mapping, we can determine the seven core journals. Use of network analysis methods analysis the social position and role ,which include the core journals, reference texts and the author. Through the keywords and Noun phrases, we draw the knowledge mapping of public health and preventive medicine, and reveal the research topic and hot from 2005?2009. I hope it can help someone who want to apply the information visualization technology in Medicine information.

Keywords: information visualization，knowledge mapping，citation analysis，cluster analysis，public health and preventive medicine