EVALUATION OF FIVE FREE MEDLINE SITES IN INTERNET
Abstract
Medline is one of the valuable sources of medical information on internet. Medline is an electronic index in which retrieval of about 11,000,000 documents from 4300 medical journals in 30 languages is possible and on the average 8000 documents are being added to it every week (1). Now, free access to Medline is possible in more than 50 sites (2-3). The quality of searching, strategy of searching and considered limitations in searching, and the interval of up-dating vary from site. In regard to these differences, researchers are faced with this question that which sites of Medline can provide more accurate and faster information. Among different free sites of Medline, five sites of BioMed Net, Dimidi, infotrieve, GratefulMed and Plumbed are the most famous ones (4-6). In the present study two indices of retrieval and precision that are more important in the evaluation of sites have been investigated. Retrieval is the number of searched documents a part from the relation of document with subject (3-5). As previous studies have proved the effect of searching language on retrieval and precision (3-6), in this study searching was performed by natural language and control vocabulary searching.
This cross-sectional study was carried out on five famous free sites of Medline on 2000. To evaluate these sites, retrieval and precision were determined and in order to control the probable effect of searching language evaluation was done by two natural language and control vocabulary searching. 36 researchers referring to the research information in Medline were selected and each of their re-question subjects (36 subjects) were searched in the five understudied sites once by the key words suggested by the researchers (natural language searching) and once by the key words determined based on subject (control vocabulary searching). The period of the searching was from 1990 to 2000. In control vocabulary searching, key words of researching subject were determined by using Mesh page in PubMed. Then searched documents were evaluated for their relation to the aimed subjects. Precision was calculated by dividing the number of related documents to the total searched documents. Data analysis was done by using paired Hest and Repeated ANOVA in SPSS soft ware.
Evaluation of Researchers of Searched Documents in five Understudied Sites was shown in table. Based on the results of this study in both natural language and control vocabulary searching there was no significant difference among the five understudied sites in regard to the retrieval and precision. These results are in line with the results of Bonham (1988) study (3). Retrieval and precision of all five sites were significantly higher in nature language searching in comparison to control vocabulary searching (P < 0/05).
This cross-sectional study was carried out on five famous free sites of Medline on 2000. To evaluate these sites, retrieval and precision were determined and in order to control the probable effect of searching language evaluation was done by two natural language and control vocabulary searching. 36 researchers referring to the research information in Medline were selected and each of their re-question subjects (36 subjects) were searched in the five understudied sites once by the key words suggested by the researchers (natural language searching) and once by the key words determined based on subject (control vocabulary searching). The period of the searching was from 1990 to 2000. In control vocabulary searching, key words of researching subject were determined by using Mesh page in PubMed. Then searched documents were evaluated for their relation to the aimed subjects. Precision was calculated by dividing the number of related documents to the total searched documents. Data analysis was done by using paired Hest and Repeated ANOVA in SPSS soft ware.
Evaluation of Researchers of Searched Documents in five Understudied Sites was shown in table. Based on the results of this study in both natural language and control vocabulary searching there was no significant difference among the five understudied sites in regard to the retrieval and precision. These results are in line with the results of Bonham (1988) study (3). Retrieval and precision of all five sites were significantly higher in nature language searching in comparison to control vocabulary searching (P < 0/05).