Publications by Dr. Perryman
Permanent URI for this collectionhttps://hdl.handle.net/11274/11193
Browse
Browsing Publications by Dr. Perryman by Subject "Bibliometrics"
Now showing 1 - 3 of 3
- Results Per Page
- Sort Options
Item Does a social network based model of journal metrics improve ranking?(University of Alberta, 2007) Perryman, Carol L.A Review of: Bollen, J., Van de Sompel, H., Smith, J.A., & Luce, R. (2005). Toward alternative metrics of journal impact: A comparison of download and citation data. Information Processing and Management, 41:1419-1440. Objective: To test a new model for measuring journal impact by using principles of social networking. Research questions are as follows: Can valid networks of journal relationships be derived from reader article download patterns registered in a digital library’s server logs? Can social network metrics of journal impact validly be calculated from the structure of such networks? If so, how do the resulting journal impact rankings relate to the ISI impact factor (IF)? Design: Bibliometric, social network centrality analysis Setting: Los Alamos National Laboratory (LANL), New Mexico Subjects: 40,847 full-text articles downloaded from a large digital library by 1,858 unique users over a 6 month period. Methods: Full-text article downloads from a large digital library for a six-month period were examined using social networking analysis methods. ISSNs for journals in which the retrieved articles were published were paired based upon the proximity of use by the same user, based on the supposition that proximal downloads are related in some way. Reader-Generated Networks (RGNs) were then tested for small-world characteristics. The resulting RGN data were then compared with Author-Generated Networks (AGNs) for the same journals indexed in the Institute of Scientific Information (ISI) annual impact factor (IF) rankings, in the Journal Citation Reports (JCR) database. Next, a sample of the AGN-derived pairings was examined by a team of 22 scientists, who were asked to rate the strength of relationships between journals on a five-point scale. Centrality ratings were calculated for the AGN and RGN sets of journals, as well as for the ISI IF. Main results: Closeness and centrality rankings for the ISI IF and the AGN metrics were low, but significant, suggesting that centrality metrics are an acceptable impact metric. Comparison between the RGN and ISI IF data found marked differences, with RGN mirroring local population needs to a much higher degree, and with a non-significant correlation between the ISI IF and RGN ranking, while AGN and RGN centrality rankings show significant centrality and closeness and betweenness correlations. RGN network ranking identified highly localized foci of interest for the LANL, as well as “interest-bridging” subject areas pointing to possible emerging interests among the scientists. Conclusion: The study results appear to successfully demonstrate an alternative to existing journal impact ranking that can more validly and accurately reflect the practices of a local community. The authors suggest that the social network-derived methodology for identification of impact rankings avoids biases intrinsic to ISI IF as a result of frequentist metrics collected from a global user group. Although the authors resist the idea of generalizability due to the local nature of their data, they suggest that the methodology can be successfully used in other settings, and for a more global community. Finally, the authors propose the automated creation of an open-source RGN whose data could be localized for smaller communities, with potentially large implications for the existing publishing industry.Item Thematic categorization and analysis of peer reviewed articles in the LISA database, 2004-2005(University of Alberta, 2009) Perryman, Carol L.A Review of: Gonzalez-Alcaide, Gregorio, Lourdes Castello-Cogolles, Carolina Navarro-Molina, et al. “Library and Information Science Research Areas: Analysis of Journal Articles in LISA.” Journal of the American Society for Information Science and Technology 59.1 (2008): 150-4. Objective – To provide an updated categorization of Library and Information Science (LIS) publications and to identify trends in LIS research. Design – Bibliometric study. Setting – The Library and Information Science Abstracts (LISA) database via the CSA Illumina interface. Subjects – 11,273 item records published from 2004-2005 and indexed in LISA. Methods – First, a search was set up to retrieve all records from 2004-2005, limited to peer review items (called “arbitrated works” by the authors (150)) and excluding book reviews. Second, thematic descriptor terms used for the records were identified. Frequency counts for descriptor term occurrence were compiled using Microsoft Access and Pajek software programs. From the results of this search, the top terms were analyzed using the Kamada-Kawai algorithm in order to eliminate descriptor term co-occurrence frequencies under 30. A cluster analysis was used to depict thematic foci for the remaining records, providing a co-word network that visually identified topic areas of most frequent publication. Conclusions were drawn from these findings, and recommendations for further research were provided. Main Results – The authors identified 18 “thematic research core fields” (152) clustered around three large categories, “World Wide Web”, “Education”, and “Libraries”, plus 12 additional peripheral categories, and provided a schematic of field interrelationships. Conclusion – Domains of greatest focus for research “continue to be of practical and applied nature,” (153) but include increased emphasis on the World Wide Web and communications technologies, as well as on user studies. A table of the most frequently occurring areas of research along with their top three descriptor terms is provided (Table 1, 152) (e.g., “World Wide Web” as the top area of research, with “online information retrieval” (268 occurrences), “searching” (132 occurrences), and “web sites” (115 occurrences)).Item Tracking theory building and use trends in selected LIS journals: More research is needed(University of Alberta, 2007) Perryman, Carol L.Objective - The authors measure theory incidents occurring in four LIS journals between 1984-2003 in order to examine their number and quality and to analyze them by topic. A third objective, only identified later in the text of the study, was to compare theory development and use between Korean and international journals. Research questions asked include whether LIS has its own theoretical base as a discipline, and what characteristics the theoretical framework has. Design – Bibliometric study. Setting – Journal issues selected from four LIS journals for the time span from 1984 - 2003. Subjects – Two international journals, Journal of the American Society for Information Science and Technology (JASIST) and Library and Information Science Research (LISR) were selected based on their high ranking in the Social Sciences Citation Index (SSCI) impact factors. Two Korean journals, Journal of the Korean Society for Information Management (JKSIM) and Journal of the Korean Society for Library and Information Science (JKSLIS) were selected. Methods - After having determined a definition of theory, and identifying different levels of theory, the authors set up rules for the identification of theory incidents, which are defined as “events in which the author contributed to the development or the use of theory in his/her own paper” (550). Content analysis of 1661 research articles was performed to measure theory incidents according to working definitions. Interrater reliability was ensured by conducting independent coding for “subfield classification, identification of theory incidents, and quality measurement” (555), using a sample of 199 articles (random selection not specified), achieving 94-97% interrater reliability. Incidents, once identified, were evaluated for quality using Dubin’s “efficiency of law” criteria, involving measures of relatedness, directionality, co-variation, rate of change, and “profundity,” defined as the depth to which theory is incorporated into the research study. Main Results - 21.79% (n=362) of the articles contained theory incidents that were analyzed and evaluated. Trend measurement indicated an overall increase, although a slight decrease was shown in the year range 1993-2003. International journals accounted for 61.33% of theory incidents, compared to 38.67% for the Korean journals. T-testing showed that differences in means between Korean and international journals were not statistically significant. Topical theory areas were ranked by frequency. The top five areas were shown to be nearly identical between Korean and international journals. ANOVA was performed with significant results in the difference between efficiency ratings. Conclusion – The authors find that the overall proportion of theory incidents including both theory development and use increased through the 20-year time span examined, and that LIS has established its own theoretical framework based upon the frequency of incidents.