| dc.description.abstract | 由於學術期刊的數量不斷地快速增加與變化,各期刊與論文對於各系所或主題的衝擊與影響程度也因而隨著時間而不斷地在異動。然而學者專家發現要經常性地與精確性地更新其系所或研究主題的核心期刊/論文排序清單是一件相當具挑戰與繁瑣的工作,因此要如何經常、快速與正確地評估大量的期刊與文章,以產生具排序性且符合系所與研究主題的核心期刊與論文,即成為教師、研究人員或圖書館員們相當具有挑戰性的任務。因此,本研究設計與實作一個電腦輔助文獻計量系統,簡稱CABS,以輔助上述任務之達成,而此系統共包括五個子系統,分別為系所期刊文獻計量系統(DJCABS),主題期刊文獻計量系統(SJCABS),系所文章文獻計量系統(DACABS),主題文章文獻計量系統(SACABS)與生醫期刊文獻計量系統(BJCABS)。
在系所期刊文獻計量系統(DJCABS)中,JCDF與LibJF兩個方法被提出,本研究並挖掘出一個隱藏規則,其普遍存在十個樣本期刊與六個科系中,即TP-to-NoJournal比值會維持在0.07左右。在藉由重疊率,標準差距離與相關係數比較其它相關研究的二個方法,JCDF與LibJF顯示出其比專家問卷法或同質性圖書館訂購法尤佳。
在主題期刊文獻計量系統(SJCABS)中,SJCDF指標被提出,並挖掘出兩個引文規則,其中TA比值在七個樣本主題與十個樣本期刊中都會維持在0.07左右;而期刊引用比值分佈可以進一步歸納為兩個型態,第一型態為“1:1.5n:1.8n”,第二型態為“ ”,這些比值規則將有助於決定主題核心期刊區域。在藉由相關係數來與其它方法(SIM, DIF和WSM)比較後,SJCDF顯示可以與DIF及WSM達到高達0.8的相關係數值;本研究並提供數據說明期刊自我引用率所造成的嚴重雜訊,SJCDF移除此雜訊,而SIM,DIF和WSM的方法並無移除此雜訊值。
在系所文章文獻計量系統(DACABS)中,RCC,TCC,PI及CH四個指標被提出,而此四個指標乃針對不同的使用者需求而設計;本研究發現的轉折點(TP)規則為TP在四個系所及十個期刊樣本中皆座落於第4個區間,此結果為當使用TCC指標的狀況之下;而當使用RCC指標時,TP規則可以分為兩型:1)型I為TP是0.07, 2)型II為TP是0.14。而在藉由相關係數比較四個指標後,RCC與TCC各自獲得與其延伸的指標RPI/RCH與TPI/TCH之相關係數值為0.5與0.9。
在主題文章文獻計量系統(SACABS)中,SRCC, STCC, SPI與SCH四個指標被使用來產生主題的論文排序清單;在STCC指標的分析中,TP規則為在各主題皆座落於第3個區間,TC比值則皆固定於0.2,而TP的夾角則皆於70度左右;在SRCC指標的分析中,TP規則為在各主題皆座落於0.07,而TC比值與TP夾角則皆與STCC指標的結果雷同。此外,SRCC與STCC皆得到與SRPI/SRCH和STPI/STCH超過0.7與0.9的相關係數值;而其中SRCC/SRPI/SRCH得到與”Google Scholar”超過0.7的相關係數值。 | zh_TW |
| dc.description.abstract | Due to the tremendous increase and variation in serial publications, the impact of every journal to multidisciplinary departments or subjects is becoming more changeable. While scholars are finding it impossible to update their departmental/subject core journal/article ranking lists regularly and accurately. The evaluation of serial journals/papers for ranking departmental/subject core journal/paper lists becomes a very challenging task for departmental faculties and librarians. Therefore, a CABS (Computer-Aided Bibliometric Systems) was proposed. There are five subsystems in the CABS, which are the DJCABS (Departmental Journal CABS), SJCABS (Subject Journal CABS), DACABS (Departmental Article CABS), BJCABS (Biomedical Journal CABS) and SACABS (Subject Article CABS).
In the DJCABS, two methods (JCDF and LibJF) were proposed. One citation pattern was found and the ratio of TP-to-NoJournal was always around 0.07 among the 10 journals and 6 departments. After comparing with four methods via overlapping rate, standard deviation distances and correlation factor, the two proposed methods were shown to outperform the questionnaire and library subscription method.
For the SJCABS, Subtract Self-Journal Cited Factor (SJCDF) was proposed. The TA-Ratio was always around 0.07 for 7 subjects and the top 10 journals. Two types of ratios exist in the journal citation ratio distribution. Type I is “1:1.5n:1.8n” and Type II is “ ”. These ratios can be helpful when deciding the core journal area. After comparing with three other methods (SIM, DIF and WSM) via the Correlation factor, SJCDF was shown to be at an acceptance level. The journal’s self citation problem was shown to be a serious bias in this study. SJCDF removes this noise which others ignore.
As for the DACABS, four indicators (RCC, TCC, PI and CH) were proposed. These four methods were designed to satisfy different audiences’ requirements. All TP were located at the 4th segment for all departments/journals by the TCC method. Through the RCC method, TPs from different departments/journals were classified into two types. The TP site of Type I was 0.07 and Type II was 0.14. After comparing these four methods via the Correlation factor, both RCC and TCC obtained more than 0.5 and 0.9 Correlation factors with their own extended methods.
For the SACABS, four indicators (SRCC, STCC, SPI and SCH) were proposed. All TPs were located at the 3rd segment for all subjects by the STCC. The TC Ratios are all about 0.2. The TP Angles are all about 70 degrees. Using the SRCC, all TPs from different subjects were all located at the 1st segment. Both TC Ratios and TP Angles are close to the experimental results from STCC indicator. The patterns of TC Ratios and TP Angles are 0.2 and 70. In addition, both SRCC and STCC can get more than 0.7 and 0.9 correlation factors with SRPI/SRCH and STPI/STCH. SRCC/SRPI/SRCH has more than 0.72 correlation factors with ‘Google Scholar’. | en_US |