• 沒有找到結果。

期刊影響力指標探析

N/A
N/A
Protected

Academic year: 2021

Share "期刊影響力指標探析"

Copied!
21
0
0

加載中.... (立即查看全文)

全文

(1)教育資料與圖書館學 Journal of Educational Media & Library Sciences http://joemls.tku.edu.tw. Vol. 49 , no. 2 (Winter 2011) : 195-214. 期刊影響力指標探析 An Overview of the Journal Impact Indicators. 蔡 明 月 Ming-Yueh Tsay Professor E-mail: mytsay@nccu.edu.tw. English Abstract & Summary see link at the end of this article. http://joemls.tku.edu.tw.

(2) 教育資料與圖書館學 49 : 2 (Winter 2011) : 195-214. 研究論文. 期刊影響力指標探析* 蔡明月 教授 政治大學圖書資訊與檔案學研究所 E-mail: mytsay@nccu.edu.tw. 摘要 引文索引資料庫中期刊引用數據之計量指標是學術評鑑最常使用之 準則 ,本文首先以 Thomson Reuters 建置為學術界與圖書資訊服務界 廣為使用之 JCR(Journal Citation Reports)為基礎 ,並旁及在網路化 與數位化環境下 ,所產生的引文索引資料庫與系統 ,計有 Elsevier 的 Scopus 系統與西班牙 SCImago Group。針對其所開發之指標內容 及其應用與限制 ,加以探討 ,進而希冀提供學術評鑑單位 、圖書 館 、期刊出版者 、資料庫製作者 、作者與使用者之參考。 關鍵詞: 影響力指標,期刊計量,JCR,引文分析. 前 言 期刊論文為學術傳播最久遠且最普遍的管道 ,且是學術評鑑的重要指標。 期刊文獻是索引摘要等書目資料庫收錄的最主要資料來源。此外 ,期刊亦是學 術圖書館最重要的館藏資料之一 ,佔圖書館大部分的採購經費。 引用文獻分析長久以來是圖書館評估期刊品質的一種方法。隨著學術期 刊及論文的大量增加 ,研究人員投稿論文常以引文索引資料庫做為評估期刊的 依據。期刊若能為其收錄不但具有權威性 ,更代表著作在其研究領域內受到相 當的重視。國內研究機構在進行個人研究成果評估及核定研究計畫 ,各大學及 學術單位考核教師與研究人員升等及聘任時 ,亦多以發表文獻之期刊是否為 科學引文索引資料庫(Science Citation Index, SCI)、社會科學引文索引資料庫. (Social Sciences Citation Index, SSCI)、藝術與人文學引文索引資料庫(Arts and. Humanities Citation Index, AHCI)所收錄 ,以及該期刊在 JCR(Journal Citation Reports, JCR)中引用數據指標做為判定期刊優劣的參考。然而 ,採用引用數據. 指標來進行期刊評鑑 ,必須先掌握引用文獻索引指標的意義與應用 ,以免誤用 而造成不當後果。 ∗本 文部分內容曾發表於《 教育資料與圖書館學 》40週年國際學術研討會,2011年 http://joemls.tku.edu.tw 3月7-8日。 2011/07/01投稿;2011/10/29修訂;2011/11/03接受.

(3) 196. 教育資料與圖書館學 49 : 2 (Winter 2011). 自 1960 年代 Eugene Garfield 創立科學資訊研究院(Institute of Scientific. Information, ISI)以來 ,JCR 一直是提供期刊引用數據以評估期刊的唯一工具。 及至近一 、二十年 ,在網路化與數位化環境下 ,方便且加速了引文數據的處 理 ,產生許多引文資料庫與系統 ,相對的促進了大量且多元的期刊引用數據之 出現 ,也製造出不少有別於傳統的新式期刊影響力指標。 以引用或被引用為指標的評鑑方法歷年來引發不少議論 ,其中尤以自我引 用 、引用錯誤 、引用均等及二次引用等最受爭議。自我引用是引用文獻分析研 究中最為人詬病之處 ,藉由大量自我引用 ,作者 、期刊 、機構與國家膨脹其 被引用次數並誇大其貢獻。引用錯誤是引用文獻分析研究的第二大問題。即使 有編輯依標準書目針對文章的引用書目加以修正 ,引用錯誤率 ,不論是引用格 式或引用內容 ,仍居高不下 ,以致無法証明引用文獻分析的有效性(註 1)。引 用文獻分析研究的第三大爭議是「引用均等」,亦即不論被引用內容的質與量 , 視所有的引用同等重要。再且 ,同一篇作品被多次引用亦只計算一次。此種認 知 ,有所疏失 ,忽略了其他面向 ,因一篇文獻包含許多部分 ,例如:導論 、 方法論 、結果與結論等 ,引用文獻出現在那一部分 ,足以彰顯其不同的重要 性。引用文獻分析之另一重大疑慮是二次引用 ,作者並未真正參考原始文獻 , 其引用書目資料乃取自於別的出版品的參考書目(註 2)。 以引用或被引用為指標的評鑑方法歷年來雖迭有爭議 ,然因其具無感式 、 定量化及客觀化的特質 ,故作為期刊與期刊之間的相對研究仍具有不可抹殺 的正面價值。本文首先以 JCR 為基礎 ,並旁及 Elsevier 的 Scopus 系統 ,再輔以 西班牙 SCImago Group 等單位所開發之指標 ,針對其內容與使用限制 ,加以探 討。此外 ,並補充其他期刊影響力指標 ,進而希望提供學術評鑑單位 、圖書 館 、期刊出版者 、資料庫製作者 、作者與使用者之參考。. 二、JCR內容 目前存在歷史最久 、建置最完整的期刊引用數據資料庫 ,則非 Thomson. Reuters 公司所製作的 JCR 資料庫莫屬。JCR Science Edition 提供科學 170 多個領 域 6,600 多種期刊的引文分析資訊。至於 Social Sciences Edition 則提供社會科學 50 多個學科領域近 2,000 種期刊的引文分析資訊。. Thomson Reuters 公司將 WoS 收錄的期刊在 SCIE、SSCI 以及 A&HCI 三種資. 料庫中 ,被引用的數據加以分析整理 ,製作出 JCR 資料庫 ,主要可檢索下列期 刊資料:期刊被引用次數(total citation)、期刊基本資料(source data)、二年期. 刊影響係數(impact factor)、期刊即時引用指數(immediacy index)、期刊被引 用半衰期(cited half-life)、期刊引用半衰期(citing half-life)。另外 ,還可透過. 引用與被引用數據計算出二種期刊自我引用率(self-citing 與 self-cited)。以下針 對前述指標加以敘述。. http://joemls.tku.edu.tw.

(4) 197. 蔡明月:期刊影響力指標探析. ㈠ 期刊被引用次數(total citation) 期刊被引用次數 ,指期刊自創刊以來所刊登的全部論文在統計當年被引用 的總次數。. ㈡ 期刊基本資料(source data) 期刊基本資料揭示 ,該期刊於某特定年共出版之文章篇數 、平均每篇文章 附註參考書目數。全年出版文章篇數中 ,評述文章(review paper)篇數 、非評 述文章篇數 、個別文章平均列舉參考書目數。JCR 鑑別評述文章的依據採下列 四個準則:1. 文章包含一百個以上之參考書目;2. 文章出現在評述期刊或一般 期刊的評述專欄;3. 文章題名含評述(review)或綜論(overview)字樣;4. 文章 綱要提及其屬於評述或調查(survey)性質。綜而言之 ,該部分提供的資訊足以 顯示:期刊的年產量為何?那些期刊出版評述文章及出版多少?再且 ,從各期 刊平均每篇文獻包含參考書目的變化 ,亦可評估其是否具有研究價值。設若一 期刊之引用參考書目率過低 ,其學術品質自然令人懷疑。. ㈢ 二年期刊影響係數(impact factor) 期刊影響係數指示期刊每篇文章平均被引用次數。單純依被引用次數評估 期刊價值 ,利於刊期較短且每期刊載較多文章的期刊 ,例如:Nature 每年出版 945 篇 ,而 Annual Review of Biochemistry 每年出版約 20 篇。很明顯的 ,Nature 被引用機會自然較多。因此 ,JCR 設計出影響係數以修正此種差異。JCR 對影 響係數定義為:某期刊二年前出版文章總數在某一特定年平均被引用次數。換 言之 ,即第一年與第二年出版文章在第三年被引用的總數 ,除以第一年與第二 年出版文章的總數。以表1的2009年Nature為例 ,2007年出版文章數為841篇 , 2008 年 899 篇 ,兩年共計 1,740 篇。2007 年出版文章被引用次數為 31,100 次 ,. 2008 年為 28,896 次 ,二年被引用次數共計 59,996 次。因此 ,影響係數的計算法 為 59996/1740=34.48。原則上 ,影響係數值越大 ,表示該期刊影響力越大。 表 1 JCR 影響係數計算法. Journal: NATURE Impact Factor: 34.48 Cites in 2009 to articles published in: 2007 = 31100. 2008 = 28896. 07 + 08 = 59996 Number of articles published in: 2007 = 841 2008 = 899. 07 + 08 = 1740 Calculation: Cites to recent articles / Number of recent articles = 59996 / 1740 = 34.48 資料來源:檢索自 2009 年 JCR 資料庫. http://joemls.tku.edu.tw.

(5) 198. 教育資料與圖書館學 49 : 2 (Winter 2011). 雖然影響係數可修正被引用的絕對次數 ,但如此一來 ,相對地卻會偏 袒刊期長且文章少的期刊。例如:查 2009 年版 JCR,發現 Journal of Clinical. Investigation 的影響係數為 15.387,而 Pharmacological Reviews 為 17.000。看 起來後者影響力大於前者甚多 ,理應較具影饗力;實際上 Journal of Clinical. Investigation 的被引用次數(87,133 次)是 Pharmacological Reviews(8,937 次)的 9.8 倍。因此 ,以影響係數作為期刊影響力之絕對指標仍值得深思。. 總之 ,引用資料的比較只適用於同一學科領域的期刊。Arunachalam 與 Manorama 研究發現 ,醫學期刊的平均影響係數為 8.633,有機化學為 2.048。即 使在同一學科範圍 ,不同國家出版的期刊影響係數亦不相同 ,例如:醫學期刊 之平均影響係數雖為 8.633,以色列出版者為 0.350,印度則為 0.287(註 3)。. ㈣ 期刊即時引用指數(immediacy index). JCR 定義即時引用指數為:某期刊當年出版的文章在同一年的平均被引用. 次數。以表 2 2009 年 Nature 為例 ,該期刊在 2009 年出版 866 篇文章 ,這些文章. 在 2009 年被引用 7,109 次 ,故其即時引用指數為 7109/866 =8.209。 表 2 JCR 即時引用指數計算法. Journal: NATURE Immediacy Index: 8.209 Cites in 1999 to articles published in 2009 = 7109 Number of articles published in 2009 = 866 Calculation: Cites to current articles / Number of current articles = 7109 / 866 = 8.209 資料來源:檢索自 2009 年 JCR 資料庫. 即時引用指數乃期刊「即時」影響力的指標。換言之 ,即在評估一期刊平 均每篇文章多「快」被引用。此可作為查尋那一種期刊最近出版較熱門文章的根 據 ,亦即當年最搶手的期刊為何?即時引用指數的計算對較早出版較多文章的 期刊較佔優勢 ,因在一年中較早出版的期刊被引用機會多於較晚出版的期刊。 即時引用指數與影響係數最大不同在時間的分界。前者強調當年 ,後者則 為前二年的累積。二者差異在於一個顯示現今當下的影響力 ,一個觀測長久持 續的影響力。若根據 Price 的即時效應(immediacy effect)理論:「通常文獻在出 版第二 、三年被引用達到最高峰」(註 4),那麼 ,JCR 影響係數的計算以二年為 基礎似乎是有強力的根據。奇妙的是 ,Tomer 的研究卻認為高影響係數期刊亦 擁有高即時引用指數(註 5)。. (cited half-life) ㈤ 期刊被引用半衰期 期刊被引用半衰期所指為:從最近一年算起 ,一期刊被引用次數達到全部 被引用次數50%時所需之時間。以2009年Nature為例 ,如表3所示 ,加以說明。 http://joemls.tku.edu.tw 2009 年時 ,Nature 的歷年被引用累積百分比及年紀分佈如下:.

(6) 199. 蔡明月:期刊影響力指標探析. 表 3 Nature 歷年被引用累積百分比及年紀 年代 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000. 累積 % 1.47 7.45 13.89 20.57 27.61 33.11 38.96 44.77 50.82 56.91. 年紀 * 1 2 3 4 5 6 7 8 9 10. * 通常期刊一出版即計一歲 資料來源:檢索自 2009 年 JCR 資料庫. 由表 3 可明顯看出 50% 的比率是介於 8(44.77%)與 9(50.82%)年之間。由. 線性內插可得其半衰期為 8+(50–44.77/50.82–44.77)= 8 + 0.9 = 8.9 年。Nature 的. 被引用半衰期為 8.9 年 ,表示半數以上文章在出版後 8.9 年即被引用達到一半 ,. 約 17.8 年後被引用就很少很少。例如:Scientist 的半衰期為 4.4 年 ,則其被引用 的生命期限約為 8.8 年。由此可見 ,Scientist 的文章較 Nature 文章老化得快。. 期刊被引用半衰期有兩項規定:1. 期刊必須被引用 100 次以上方能列出;. 2. 當 10 年內被引次數未達全部被引次數一半時 ,其半衰期以 >10.0 表示。. ㈥ 期刊引用半衰期(citing half-life). JCR 期刊引用半衰期定義為:從最近一年算起 ,一期刊引用其他期刊次數. 達到全部引用次數 50% 時所需之時間 ,其計算方式與被引用半衰期相同。引用 半衰期長短反映引用資料的新穎與老舊。 Pravdic 與 Pekorari 認為簡訊期刊 ,如 Physics Letters、Physical Review Letters. 等之引用半衰期有較短的趨勢(註 6)。Arunachalam 與 Manorama 發現 ,開發中. 國家出版之期刊其引用文獻通常較已開發國家來得老舊 ,例如:在環境科學方 面 ,印度期刊之引用文獻中有 53% 是 10 年以上 ,只有 14% 是等於或小於四年。 相對地 ,高度開發國家出版之期刊其引用文獻超過 10 年的只佔 32%,等於或小 於四年卻佔 27%(註 7)。. ㈦ 期刊自我引用率(self-citing rate). 期刊自我引用率可由引用期刊(Citing Journal Listing)視窗檢索取得 ,該. 部分在說明某期刊引用其他期刊的情況。如表 4 所示 ,2009 年 Nature 乃引用. 期刊(citing journal),其下為在某特定年(2009)該期刊引用其他期刊的總次 數(36,789)及各年出版之期刊最近十年引用次數分佈(2009 年出版之期刊引 用 2,454 次 ,2008 年引用 4,784 次…),與其餘的十年前總引用次數(9,175)。 http://joemls.tku.edu.tw. 引用期刊之下列出被引用的期刊(cited journal),例如:Science;及其被引用.

(7) 200. 教育資料與圖書館學 49 : 2 (Winter 2011). 總數(2457)與歷年被引用次數的分佈(2009 年版被引用 110 次;2008 年版被引. 用 322 次…)。被引用期刊依被引用次數多寡排序。由表 4 可見最高被引用期刊. 為引用期刊本身。此說明當該期刊為引用期刊時 ,其自我引用率為 4278/36789. (11.6%),高於引用其他期刊的比率。 表 4 JCR 引用期刊表. Citing Journal Listing(Citing Journal = NATURE) < ---Number of times articles published this year were cited in 2009--- > Impact Cited factor journal All 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 Rest All Journals 36789 2454 4784 4382 3662 2976 2509 2066 1848 1534 1399 9175 34.480 NATURE 4278 672 591 497 349 343 264 211 188 181 149 833 29.747 SCIENCE 2457 110 322 302 241 207 182 117 158 131 121 566 9.432 P NATL ACAD 1574 79 232 203 157 156 111 108 67 72 57 332 SCI USA 31.152 CELL 1264 74 135 174 176 132 67 56 66 46 44 294 7.364 ASTROPHYS J 811 65 104 87 74 63 52 47 29 24 45 221 5.328 J BIOL CHEM 737 10 57 61 71 54 73 57 50 50 43 211 7.328 PHYS REV LETT 706 41 103 89 69 47 39 27 31 13 25 222 34.284 NAT GENET 415 36 76 60 62 32 16 19 19 14 16 65. ㈧ 期刊自我被引用率(self-cited rate). 資料來源:檢索自 2009 年 JCR 資料庫. 期 刊 自 我 被 引 用 率 可 由 被 引 用 期 刊(cited journal listing)視 窗 檢 索 取. 得 ,表 5 說明某期刊被其他期刊引用情況 ,其中 2009 年 Nature 乃被引用期. 刊(cited journal),其下為在某特定年(2009)該期刊被其他期刊引用的總次. 數(483,039),及各年出版期刊最近十年的被引用次數分佈(2009 年 7,109 次 , 2008 年 28,896 次…),與其餘十年前的加被引用總次數(208,136)。被引用期刊. 之下依引用次數列出引用期刊(citing journal),例如:Science;每一引用期刊 亦伴隨有某特定年的引用總次數(2,279)與歷年引用次數分佈(2009 年版引用 125 次 ,2008 年版引用 295 次 ,2007 年版引用 259 次…)。值得一提的是 ,當 表 5 JCR 被引用期刊表. Cited Journal Listing(Cited Journal = NATURE) < ---Number of times articles published this year were cited in 2009--- > Impact Citing factor journal All 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 Rest All Journals 483039 7109 2889 31100 32267 33977 26608 28215 28097 29228 29406208136 4.351 PLOS ONE 8257 122 548 723 681 756 516 513 521 484 520 2873 9.432 P NATL ACAD 7112 182 622 684 609 589 411 447 455 382 353 2378 SCI USA 5.328 J BIOL CHEM 5799 86 337 406 403 381 286 354 365 303 340 2538 3.475 PHYS REV B 5457 132 738 434 476 525 466 409 381 324 294 1278 34.480 NATURE 4278 672 591 497 349 343 264 211 188 181 149 833 7.328 PHYS REV 3507 83 499 325 361 324 338 261 195 239 185 697 LETT 7.178 J NEUROSCI 3092 37 147 144 233 164 145 160 212 161 167 1522 7.364 ASTROPHYS J 2826 81 187 171 362 353 111 139 90 78 38 1216 29.747 SCIENCE 2279 125 295 259 200 211 162 143 113 94 113 564. http://joemls.tku.edu.tw. 資料來源:檢索自 2009 年 JCR 資料庫.

(8) 201. 蔡明月:期刊影響力指標探析. Nature 為被引用期刊時 ,自我被引用(self-cited)率為 4278/483039(0.01%)。總 之 ,期刊間「互引」的連繫 ,說明彼此關係之遠近親疏 ,亦不啻為描述期刊主 題網絡結構之理想工具。 期刊自我引用率是評鑑期刊穩定性與連續性的指標 ,自我引用率高說明該 期刊刊登的文章具相關性或延續性 ,在其領域形成獨特的風格與影響力 ,然亦 反映其封閉性與其他期刊的交流不足。 總結上述 ,JCR 提供之期刊影響力指標計有:1. 期刊基本資料 ,2. 期刊被 引用次數 ,3. 二年期刊影響係數 ,4. 期刊即時引用指數 ,5. 期刊自我引用率 , 6. 期刊自我被引用率 ,7. 期刊引用半衰期 ,8. 期刊被引用半衰期。. 三、JCR之限制 JCR 是 ISI 有效應用電腦科技製作 SCI 後的附加產物 ,雖其成果獲得不少喝 采 ,但仍有其不足之處 ,造成使用時有所限制。茲略述如下: ㈠引用次數固可顯示期刊被引用情形及其價值 ,然高度引用並不代表高度 使用。例如:新知報導性質期刊 ,如:Scientific American 與 New Scientist 被閱 讀機會很多 ,但被引用排名卻很低 ,這並不意謂它們較少被使用或較不重要。 ㈡ JCR 期刊指標中最常被用來評鑑期刊等級的是影響係數 ,原則上 ,IF. 越大表示該期刊影響較大。因 IF 是分數的結果 ,分子與分母的關係決定其數 值的大小 ,所以期刊的性質會影響其影響係數。一般而言 ,評述性質的期刊 (review journal),例如:Annual Review of Information Science and Technology、 MIS Quarterly,刊期較長 ,常見為季刊 、半年刊或年刊 ,刊載文章較長 ,文章 篇數較少。這類期刊文章是進行文獻探討工作時 ,最常被使用的文獻類型 ,被 引用機會相當高。如此一來 ,在分子大分母小的情況下 ,IF 自然較高。 ㈢評述性質期刊的文獻多屬整理性質 ,未見學術的實證與原創性 ,讀者引 用的行為應是二次引用 ,更嚴格的說法 ,讀者應引用最原始的著作 ,才是學術 引用該探索的本質。以台灣生產的 TSSCI 收錄文章為例 ,由於社會科學引進許多 外國知識 ,其引用參考書目大多數是英文文獻;即使不是外國原文亦多為中文翻 譯 ,如此計算出的被引用次數實無法真實反映其學術的貢獻程度。 ㈣期刊 IF 每年都不一樣 ,一個每年都在變動的工具 ,如何作為一公正客觀 之指標?要問的問題是 ,要採用那一年的作為根據呢?從統計學方面看 ,觀察 15 年的變化是較有效的。換言之 ,求其 15 年的平均值再比較 ,較能看出其顯著 性。期刊品質或學術貢獻 ,需要時間的檢驗以產生累積的效果。最後 ,IF 相差 0.1 或 0.2,就能斷定誰優誰劣嗎? ㈤由於影響係數是一個平均值 ,期刊每年出版文章篇數或計算的年數均會 影響其數值。出版文章篇數越多的期刊其每年影響係數的改變較少 (例如:150 http://joemls.tku.edu.tw. 篇以上的期刊約在正負 15%,35 篇以下者在正負 40% 之間)。一個出版 140 篇.

(9) 202. 教育資料與圖書館學 49 : 2 (Winter 2011). 文章的期刊其影響係數為 1.5,與另一出版相同文章篇數的期刊但其影響係數 是 1.24,二者其實沒有什麼不同。大略估計 ,期刊之間的影響係數差異若少於. 0.25 則應屬於同一等級(註 8)。. ㈥計算時間的長度亦會影響影響係數。據研究顯示 ,JCR 以二年為計算基 準的影響係數 ,各期刊每年的數值 ,起伏變動相當大。然若改以五年為計算基 準 ,則各期刊每年的數值較為平緩穩定。 影響係數除了如上所述客觀限制外 ,其中尚隱含許多不為人所透徹了解的 錯誤認知 ,值得吾人深思。經由研究顯示 ,基礎科學或較單一主題的期刊比應 用科學或特定主題期刊之影響係數較高 ,例如:物理學平均影響係數為 1.6 較材. 料科學的影響係數 0 . 7 高( 註 9 )。E g g h e 與 R o u s s e a u 查考科學引用文獻索引 的 J C R 發現大多數的期刊影響因素都很低 ,有 9 7 % 低於 5,6 3 % 小於等於. 一( 註 10 ) ;再且 ,大多數文章很少被引用。即使在同一學科 ,期刊性質不 同或期刊中刊載不同性質的文章仍會左右影響係數。 此外 ,使用 JCR 依學科主題分類的期刊排序 ,應謹慎從事。一種期刊往往. 不只限一主題 ,在 JCR 中也就不只歸入一類而已 ,在某一學科領域排名在前的 期刊 ,在另一學科則可能排名最後 ,處理上稍不妥當則可能完全失去意義。再 且 ,直接就學科類別下的影響係數大小決定期刊的排名是相當危險的 ,因某期 刊被引用很多 ,可能是在別的領域被引用。例如:MIS Quarterly 是資管領域的 主流期刊 ,該期刊最近 12 年才被歸入圖資類 ,屬於邊緣期刊 ,在圖資類影響係. 數排名第一 ,而被引用卻大多在資管領域。因此 ,學科領域不同 ,期刊影響係 數不可互相比較。此外 ,應極力避免採用一影響係數的絕對值(absolute value) 去評量期刊品質 ,因除了受前述變數影響之外 ,不同學科領域長期發展情勢亦 影響了影響係數。若只因第九年的影響係數低於其他年代 ,就判定第九年的研 究做得不如其他年代 ,將是不合理的推斷。 Seglen 亦觀察到文章的被引用情況與期刊的影響係數之間無相關關係存 在。換言之 ,文章被引用與否與期刊聲望無關 ,一些擁有高影響係數的期刊 , 只因它們出版了一小部分高影響力的文章。此外 ,合作者多寡亦會顯示影響係 數的差異 ,當然合作者人數亦受到學科性質的影響 ,例如:社會科學類每篇文 章平均作者數為二人 ,至於生命科學則超過四人(註 11)。. 基於上述種種限制 ,無怪乎 Garfield 宣稱:「他在 1955 年首次提出 impact. factor 的觀念 ,當時並未想到會造成今日廣泛的衝突 ,impact factor 是 Current. Contents 與 SCI 資料庫以及圖書館用來選擇好期刊之用 ,然而在各種情況下被誤 用了 ,尤其是評鑑個別研究者。」(註 12) 綜上所述 ,影響係數實際上受到許多因素的操縱 ,例如:期刊的主題 、期. 刊的類型(簡訊期刊 、評述期刊)、每篇文章作者人數 、期刊所含文章篇數以及 引用次數。. http://joemls.tku.edu.tw.

(10) 203. 蔡明月:期刊影響力指標探析. 除了影響係數的探討外 ,期刊與文章的性質還會左右其他的引用數據指 標。一般而言 ,刊期短 、出版快速的簡訊期刊(letters journal),例如 Physics. Letters, Physical Review Letters 通常有較高的即時引用指數以及較短的被引用半. 衰期 ,約在二年之內。換言之 ,其老化較快。反之 ,刊載完整論文的期刊 ,其 被引用高峰在文章出版後第三年 ,被引用半衰期較長 ,且老化趨向平緩。至於 評述期刊 ,其即時引用指數相對地自然較低 ,通常在出版後數年其被引用才達 到高峰。被引用半衰期相對的加長 ,老化速度趨緩。 總之 ,JCR 使用者對引用模式必須有一通盤的了解與掌控 ,才能將 JCR 各 式各樣的期刊引用數據加以巧妙的應用 ,存在有多種變數會影響引用次數 、影 響係數與即時引用指數。舉凡作者的聲望 、學科特性 、語言 、刊期 、期刊銷售 量 、圖書館館藏情況 、期刊的可得性 、被索引摘要工具收錄情況 、研究經費優 先次序等均包括在內。例如:非英語期刊因不易取得 ,自然被全世界引用的可 能性降低。 最後 ,JCR 的期刊被引用與引用指標 ,是期刊的指標 ,期刊被引用很少 或很多 ,並不代表個人文章被引用 ,若要突顯國際被引用競爭力仍應回歸至 SCI、SSCI 與 AHCI 等 citation index 的檢索。然而 ,真要論到被引用的事實 ,還 得觀照到如本文前言所論及引文分析的諸多爭議與問題。. 四、新期刊影響力指標 誠如第三節所述 ,1990 年以前 JCR 可以說是唯一提供期刊引用數據之資料. 庫。1990 年之後加入許多商業引文系統 ,例如 Elsevier 公司之 Scopus,提供期 刊引用數據。自 1990 年代末期 ,JCR 面臨許多期刊評鑑新需求 ,陸續修訂開發. 出不少新期刊影響力指標 ,例如:期刊五年影響係數(five year impact factor)、 期刊 H index、期刊領域排名(journal rank in categories)、Eigenfactor Score、 Article InfluenceTM 等。此外 ,Elsevier 公司所創立之 Scopus 期刊分析(Journal. Analyzer)系統亦針對期刊建置了一些特別的影響力指標 ,例如:期刊未被引. 用 %(% not cited)、標準化期刊影響力(Source Normalized Impact per Paper,. SNIP)、SCImago Journal Rank(SJR)等 ,其中 Eigenfactor Score 與 SCImago. Journal Rank 是以 Google page rank 的做法為參考依據。茲分別就 JCR 與 Scopus. 新期刊影響力指標簡介如下(註 13)。. ㈠ JCR 新期刊影響力指標 1. 期刊五年影響係數( 5-Year Impact Factor) 社會科學領域的期刊需經過較久的時間 ,引用才會達到高峰。因此 ,JCR 設計出以五年為計量基礎的五年影響係數 ,作為兩年影響係數的補充與參考。 http://joemls.tku.edu.tw. 五年影響係數是指該期刊於早五年所出版的文獻數在某一特定年份被其它文章.

(11) 204. 教育資料與圖書館學 49 : 2 (Winter 2011). 引用的平均數 ,亦即【該期刊當年度前 5 年所有出版文獻被引用總次數】除以. 【該期刊前 5 年所有出版文獻總數】。其提供年代更長的期刊引用趨勢 ,引用活 動的查看範圍因而隨之放大 ,使用者更能快速瀏覽豐富的資料 ,進而可反映期 刊在較長時間的學術影響力 ,對期刊品質穩定性的評鑑較為精確。期刊五年影 響係數之計算公式與表 1 顯示之期刊二年影響係數相同 ,惟期刊被引用次數與 出版文獻篇數之數據由二年改為五年。 2. 期刊 H index. H-index 是 J.E. Hirsch 於 2005 年所發明一種新的計量指標(註 14),原是設計. 作為科學研究者個人研究產出之量化評鑑指標。JCR 將其應用於期刊之評量。 期刊 h index 是指某特定期刊有 n 篇文章 ,其中有 h 篇被引用 h 次以上。簡單來 說 ,就是將作者歷年來所有著作依照被引用次數由高至低的順序排序 ,當文章 數小於或等於被引用次數時 ,該數字即為作者的 h index。例如:作者 A 共有 18. 篇文章 ,依照被引用次數由高至低排序 ,第 10 篇文章被引用 11 次 ,第 11 篇被. 引用 10 次 ,所以 A 作者的 h index 就是 10。H index 的應用從最初的作者比較 , 已進一步發展到長期觀測機構 、國家 、期刊 、研究領域……等標的之重要指 標。 3. 期刊領域排名(journal rank in categories) 期刊領域排名可了解該期刊在各領域的排名情形。該指標係根據各領域之 期刊影響係數排序後 ,所得的名次。若該期刊屬多個領域 ,則有不同的名次。 此外 ,除實際名次外 ,亦可知該期刊於該領域的四分位百分比 ,Q1 為該領域 影響係數名次前 25%,Q2 為 25%-50%,Q3 為 50%-75%,Q4 則為後 75%。例. 如:Nature 2008 年的 impact factor 值為 31.434,其在多學科(multidisciplinary. sciences)領域 42 種期刊的影響係數排名第 1,為前 25%(Q1)。再以 Leukemia. 期刊為例 ,該期刊可分類入二個領域 ,其在 hematology 領域(總期刊數有 62 種). 之影響係數排名第 4,為前 25% (Q1)。同時在 oncology 領域(總期刊數有 141 種) 影響係數排名第 9,亦為前 25%(Q1)。 4. Eigenfactor Score. Eigenfactor Score 可了解特定期刊在學術引用社群的總影響力。利用期刊引. 用網絡(citation network)找出評估較具權威性的期刊 ,其原理類似 Google page. rank的概念。若該期刊被較具影響力之期刊引用 ,則得到較高的引文值。此外 , 具有相同引用次數的期刊 ,A 被較具影響力的期刊引用 ,B 被一般期刊引用 , 則 A 得到較高數值。其理論基礎來自於模擬研究者藉由引用行為從 A 移動至 B、C…D 期刊的隨機過程。採用近五年的 citation widow,詳細計算公式請參考. http://www.eigenfactor.org/methods.htm。Eigenfactor Score 特色為忽略期刊自我引 用;將科學 、社會科學期刊一起計算;可考量不同學科領域的引用差異 ,讓各. http://joemls.tku.edu.tw.

(12) 蔡明月:期刊影響力指標探析. 205. 學科能在同一引文基準點上進行比較;出版文章數越多的期刊可能具較高引用 次數 ,能見度較高 ,通常有較高的 Eigenfactor。 5. Article InfluenceTM. Article Influence TM 可提供該期刊平均每篇文章的影響力 ,其計算方式為. (期刊的 Eigenfactor/ 該期刊標準化五年文章數)×0.01。在 JCR 資料庫所有文. 章的平均值為 1,若該期刊 Article Influence 大於 1,表示該期刊每篇文章影響. 力居於 JCR 資料庫所有文章的平均值以上;反之 ,小於 1 則在平均值以下。以 Leukemia 期刊為例 ,其 Eigenfactor 為 0.0562,Article Influence 為 1.931。. ㈡ Scopus 新期刊影響力指標 除 JCR 之外 ,Elsevier 公司所創立之 Scopus 期刊分析系統亦針對期刊建置. 了一些影響力指標。這些指標除了與早期 JCR 相同的傳統指標 ,例如:期刊被 引用次數 、期刊出版文章篇數 ,還包括一些新指標 ,例如:期刊未被引用 %、 標準化期刊影響力 、SCImago Journal Rank(SJR)。此外 ,該系統亦針對多項指 標提供走勢圖(trend line)。茲進一步敘述 Scopus 新期刊影響力指標如下。 1. 期刊未被引用 %(% not cited). 期刊未被引用百分比顯示單一期刊年度被引用次數為零的文章篇數占總篇 數之比例。 ⑴ 標準化期刊影響力(Source Normalized Impact per Paper, SNIP) 此指標在修正跨領域被引用量的差異 ,其目的在計算期刊每篇文章平均被 引用次數 ,以顯示在其領域中的被引用潛力。 ⑵ SCImago Journal Rank(SJR) 該指標計算加上權重之後的每篇文章被引用值 ,其目的在區隔被有聲望期 刊(例如:Science 或 Nature)引用 ,其價值高於一般普通期刊。. 五、其他期刊影響力評估指標 除上述各期刊影響力指標 ,據 Lancaster 分析 ,尚可補充下列指標(註 15) :. ㈠ 期刊大小 期刊大小(size)除指期刊出版文章篇數多寡外 ,另一種衡量方法是以字 數或頁數作為計算單位。對某些決策之執行而言 ,例如:圖書館館藏空間或資 料庫容量的規劃 ,以期刊頁數作為考量 ,反而更為重要。大多數書目計量研究 是以期刊所含文章篇數作為計量單元 ,而不計其長度。然而 ,若將參考書目視 為評估學術價值的根據時 ,以文章長度作為反映參考的密度 ,則變得有意義多 了。舉例而言 ,一篇只有二頁的文章 ,包含 15 個參考書目 ,比起一篇有 50 頁 http://joemls.tku.edu.tw 的文章亦包含同樣多個參考書目 ,自然令人印象較深刻。反之 ,當考慮一期.

(13) 206. 教育資料與圖書館學 49 : 2 (Winter 2011). 刊被引用的情形時 ,期刊的文章篇數 ,顯然較文章長度來得重要。Sengupta 指 出 ,以每篇文章被引用次數而非每個字被引用次數來排列期刊名次 ,其結果是 不一樣的。JCR 提出評估期刊的要素 ,計有出版文章篇數及參考書目數量 ,但 不包括出版的字數及頁數(註 16)。總之 ,Narin 與 Carpenter 認同 ,從參考書目 總數大致可推斷文章的長度或期刊的大小(註 17)。. ㈡ 期刊訂閱量 一般而言 ,期刊被訂閱越多 ,應可推測該期刊較好 ,這當中期刊的訂費 及贈閱有必然的影響。圖書館進行期刊淘汰時 ,因期刊訂費低廉 ,雖其品質漸 差 ,讀者使用漸少 ,圖書館仍不致於費事地給予刪除。此外 ,專業學會出版之 期刊亦會主動贈送會員或相關權威機構。雖然如此 ,訂閱量仍不失為一公平且 良好之指標。若甲期刊的訂閱量為乙期刊的十倍 ,至少 ,可肯定的是甲期刊的 潛在影響力(即潛在讀者)大大超過乙期刊。Bennion 與 Karschamroon,利用多 重迴歸 ,分析各種物理學期刊之訂閱量 、出版文章篇數 、被引用次數 、影響係 數 ,及物理學家的評鑑等參數 ,結果發現期刊訂閱量有較好的預測效果。期刊 的實際使用很難決定 ,引用卻又只能反映著作時的使用 ,因此 ,期刊訂閱量不 失為一「潛在使用」的良好指標(註 18)。. ㈢ 影響力 Narin與Carpenter提出 ,以影響力(influence)來評估期刊。其作法較複雜。 因涉及某期刊與其他期刊間的互引率 ,必須用電腦加以計算始可完成。茲以 甲 、乙 、丙三種期刊舉例說明如下。. 圖1 三種期刊互引舉例 圖 1 所示 ,甲被乙引用 397 次 ,甲引用乙 612 次;甲被丙引用 85 次 ,甲引. 用丙 218 次 ,則甲期刊的引用影響力為(397+85)/(612+218)= 482/830 = 0.58。 同樣地 ,乙被甲引用 612 次 ,乙引用甲 397 次;乙被丙引用 73 次 ,乙引用丙 43. 次 ,則乙的引用影響力為(612+73)/(397+43)= 685/440 = 1.56。此外 ,丙被甲 引用 218 次 ,丙引用甲 85 次;丙被乙引用 43 次 ,丙引用乙 73 次 ,則丙的引用影. 響力為(218+43)/(85+73)= 261/158 = 1.65。由此可見丙的影響力最高(註 19) 。. 上面舉例 ,只有三種期刊之互引比較 ,實際作業時 ,每一種期刊是與資料. http://joemls.tku.edu.tw. 庫中所有其他期刊相比。除此之外 ,計算期刊的總影響值 ,尚須以期刊影響力.

(14) 207. 蔡明月:期刊影響力指標探析. 乘上每篇文章平均的參考文獻數及某期限內期刊出版之文章篇數。例如:丙期 刊平均每篇文章所含參考文獻為 14,出版文章篇數為 900,則其總影響值為 1.65 × l4 ×900 = 20,790。. 此方法因考慮到期刊的引用與其他期刊間的相互關係 ,再加上文章長度及 文章篇數 ,故較影響係數完善。然而 ,因涉及變數過多 ,執行自然較複雜且困 難。. ㈣ 其他 除上述三項其他準則外 ,Hawkins 尚建議以期刊所包含文章與學科主題的 相關程度作為評估準則。例如:甲期刊文章約 100% 是關於空氣污染的 ,乙期 刊有 80%,丙期刊只有 50%。對於專門圖書館或專門學科資料庫的建設與管理. 而言 ,此準則尤其重要(註 20)。最後 ,「成本」往往是任何評估要素中的最根 本。一切評估無非要在最經濟的條件下 ,達到最高的效益。期刊的評估準則 , 自不例外 ,因此 ,期刊的訂費與各種產出參數間的關係 ,理應一一審慎考量。. 六、結 語 引用文獻從被引用次數與內容分析文獻的品質。引用關係無時不在擴展 , 隨時可再生產 ,經由重複性的操作產生引用關係網。由動態的引用關係網研究 學科之間的關係來看 ,「引用」是學術傳播的一種功能;由評估科學政策的目的 來看 ,引用是科學系統的一種報酬指標;由引用文獻的觀點來看 ,引用則是對 另一文獻的參考 ,二者為引用與被引用的關係。引用 ,絕對是一個個人的操作 行為。誠如何光國所言;「引用文獻原理只是本位觀和價值觀二者間的論衡。價 值觀代表引用者個人對原著引用價值的一種主觀評價。價值觀人人各異 ,時時 不同 ,而且還常因周圍環境而變化。因此 ,引用文獻的應用實際上為一種受客 觀因素支配的主觀行為。」(註 21) 引用文獻分析的評價 ,正反互見 ,錯綜紛歧。引用行為 、引用動機及引用 功能複雜多變。欲正確有效實施引用文獻分析研究 ,非得對引用文獻分析的限 制 、引用文獻索引的缺失 、影響引用文獻分析研究的因素 ,及引用文獻分析所 遭遇的問題等重要課題 ,充分掌握 ,謹慎應用 ,否則無法竟其功。 學術研究的追求貴在交流傳播 ,傳播要有媒介 ,出版尤其期刊出版是最廣 泛的傳播途徑。期刊論文發表之後 ,更應關心的層面是其產生的回應 ,因此 , 評鑑時採用引文索引資料庫 ,進一步得知被引用狀況 ,將量化評鑑提升至質化 評鑑的層面 ,是一合理的做法。然而應考量的是 ,以引文索引來做評鑑時 , 不是只有 WoS 系統存在 ,目前尚有很多相關的引文索引資料庫 ,例如:Google. Scholar、Scopus、Cite Base, CiteSeer, NASA ads, PubMed Central 等。是以 ,就 國際化的能見度而言 ,評鑑應採用多樣化資料庫。. http://joemls.tku.edu.tw.

(15) 208. 教育資料與圖書館學 49 : 2 (Winter 2011). 大型期刊引用數據資料庫 ,例如:WoS 之 JCR、Scopus 與大陸 CNKI 之中. 國引文資料庫(Chinese Citation Database, CCD),是可將所有學科的期刊引用資 料放在一個平台上的工具。透過引用與被引用的機制 ,它們是觀察與檢驗學術 研究在國際上所造成影響之最直接工具。工具本身是沒有生命的 ,使用工具的 人賦予它價值。因此 ,應秉持工具理性的態度面對。 期刊引用資料庫從引用文獻的特性 ,探討引用與被引用的關係 ,進而作為 評定期刊或其他文獻的參考工具。雖工具內含有不可避免之缺失 ,尤其期刊多 變的特質 ,加上引用文獻的錯誤與含混不清 ,使其無法長期正確的掌握期刊之 間的關係。不過這些缺點 ,比起其正面貢獻 ,倒也是「瑕不掩瑜」。尤其其研究 方法與理念仍值得我們多多學習。期刊的特性本即複雜多變 ,如何選定評估其 價值的準則 ,端賴從事者依其特定需求選擇適當的準則加以施行 ,才是睿智的 作法。. 註 釋 註 1 H. F. Moed & Th. N. Van Leeuwen, “Improving the Accuracy of Institute for Scientific Information's Journal Impact Factors,” Journal of the American Society for Information Science 46, no. 6 (1995): 461-467.. 註 2 Sture Hansson, “Impact Factor as a Misleading Tool in Evaluation of Medical Journals,” Lancet 346, no. 8979 (1995): 906. 註 3 S. Arunachalam & K. Manorama, “Are Citation-Based Quantitative Techniques Adequate for Measuring Science on the Periphery?” Scientometrics 15, no. 5-6 (1989): 393-408. 註 4 Derek de Solla Price, “Citation Measures of Hard Science, Soft Science, Technology, and Non-Science,” in Communication Among Scientists and Engineers, ed. Carnot E. Nelson & Donald K. Pollock. (Lexington, MA: Heath Lexington, 1970), 9. 註 5 Christinger Tomer, “A Statistical Assessment of Two Measures of Citation: The Impact Factor and the Immediacy Index,” Information Processing and Management 22, no. 3. (1986): 251-258. 註 6 Nevenka Pravdić & R. Pekorari, “The Citing Practices of the Authors to the National Journals in Mathematics, Physics and Chemistry,” Scientometrics 8, no. 3-4 (1985): 233-246. 註 7 S. Arunachalam & K. Manorama, “Are Citation-Based Quantitative Techniques Adequate for Measuring Science on the Periphery?” Scientometrics 15, no. 5-6 (1989):. 393-408.. 註 8 Amin, M. 演講講義 “Journal Publishers and Impact Factors,” in Proceedings of the 8th International Conference on Scientometrics and Informetrics (Sydney, 16-20 July 2001), 14-15. 註 9 Ibid.. http://joemls.tku.edu.tw.

(16) 209. 蔡明月:期刊影響力指標探析. 註 10 Leo Egghe & Ronald Rousseau, “Averaging and Globalising Quotients of Informetric and Scientometric Data,” Journal of Information Science 22, no. 3 (1996): 165-170. 註 11 Per O. Seglen, “Causal Relationship between Article Citedness and Journal Impact,” Journal of the American Society for Information Science 45, no. 1(1994): 1-11. 註 12 Eugene Garfield, “The Use of JCR and JPI in measuring short and long term journal impact,” in Council of Scientific Editors Annual Meeting held in 9 May 2000, http://. www.garfield.library.upenn.edu/papers/cseimpactfactor05092000.html (accessed October 29, 2011). 註 13 碩睿資訊有限公司 ,Journal Citation Reports PPT,2010;Elsevier 教育訓練 , Scopus 資料庫 PPT,2010。 註 14 J. E. Hirsh, “An Index to Quantify an Individual’s Scientific Research Output,” Proceedings of the National Academy of Sciences of the United States of America 102, no. 46. (November 2005): 16569-16572. 註 15 F. W. Lancaster,伊利諾大學圖書館與資訊科學研究所 ,書目計量學課程講 義 ,1985 註 16 I. N. Sengupta, “Three New Parameters in Bibliometric Research and Their Application to Rerank Periodicals in the Field of Biochemistry,” Scientometrics 10, no. 5-6 (1986): 235-242. 註 17 Francis Narin & Mark P. Carpente, “National Publications and Citation Comparisons,” Journal of the American Society for Information Science 26, no. 1 (1975): 80-93.. 註 18 Bruce C. Bennion & Sunee Karschamroon, “Multivariate Regression Models for Estimating Journal Usefulness in Physics,” Journal of Documentation 40, no. 3 (1984): 217-227. 註 19 Francis Narin & Mark P. Carpente, “National Publications and Citation Comparisons,” Journal of the American Society for Information Science 26, no. 1 (1975): 80-93.. 註 20 D. T. Hawkins, “The Percentage Distribution: A Method of Ranking Journals,” Proceedings of the American Society for Information Science 16(1979): 230-235. 註 21 何光國 ,文獻計量學導論(台北市:三民書局 ,1994),187。. http://joemls.tku.edu.tw.

(17) Journal of Educational Media & Library Sciences 49 : 2 (Winter 2011) : 195-214. Research Article. An Overview of the Journal Impact Indicators* Ming-Yueh Tsay. Professor Graduate Institute of Library, Information and Archival Studies National Chengchi University Taipei, Taiwan E-mail: mytsay@nccu.edu.tw. Abstract This article presents the journal impact indicators, mainly covered in the Journal Citation Reports (JCR), Web of Science, that developed by the Institute of Scientific Information (ISI, today part of Thomson Reuters). These indicators include total publications, total citations, 2-year impact factor, 5-year impact factor, immediacy index, cited half-life, citing half-life, self-citing rate, self-cited rate, H index, Eigenfactor score, article influence score. In addition to JCR of WOS, Scopus supplies other unique journal impact indicators, i.e., % not cited, source normalized impact per paper (SNIP) and SCImago journal rank (SJR). Moreover, the applications and limitations of these indicators are also discussed. Keywords: Journal impact indicator; Journal metric; Journal Citation Reports; Citation analysis SUMMARY. Journal literatures are the most commonly used channel for scholarly communication and index for academic assessment. They are also the main source of bibliographic databases. In addition, journals are one of the most important collections for academic and research libraries and sharing a major portion of the library budget. Individual researchers often judge the value of a journal based on its inclusion by the citation index database, and then decide which journal they would submit their manuscripts to. Many academic organizations evaluate and appreciate their faculty’s research results and projects on the basis of the journals covered by the citation database and the citation data that those journals demonstrated. Among these data, journal impact factor is the most used indicator for judging the quality of the journal. The purpose of this article is to investigate journal impact indicator that established first by Thomson Reuters Journal Citation Reports (JCR) and later Scopus of the Elsevier also. The journal impact indicator in the JCR includes total citation, source data, 2-year impact factor, immediacy index, cited half-life, citing half-life, 5-year impact factor, H index, journal rank in categories, Eigenfactor score and article influence. Definitions of five new terms by JCR are given in the following: * Part of this article had been presented at The International Conference of the 40th Anniversary of Journal of Educational Media & Library Sciences, March 7-8, 2011.  http://joemls.tku.edu.tw Thomson Reuters, “Journal Citation Reports,” http://admin-apps.webofknowledge.com/ JCR/help/h_jrnlinfo.htm#jrnlrec (accessed January 6, 2012)..

(18) Ming-Yueh Tsay: An Overview of the Journal Impact Indicators. 211. 1. Five-year Impact Factor Journal impact factor relates the number of citations a journal receives to the number of articles it publishes. The 5-year journal impact factor is “the average number of times articles from the journal published in the past five years have been cited in the JCR year”. The calculation of the 5-year impact factor follows the same method of 2-year impact factor. The only difference is the time frame. It is calculated by “dividing the number of citations in the JCR year by the total number of articles published in the five previous years”. 2. H Index The h-index was established by Jorge E. Hirsch to quantify a scientist’s research output. Its original definition is “a scientist has an index h if h of his p papers has at least h citation each and the other p-h papers have fewer than h citations each”. The h index of a journal follows the same method. It is the number of articles h which have at least h citations. “The h-index counts and is based on a list of publications ranked in descending order by the times cited count”. For example, an h-index of 30 means there are 30 items that have 30 citations or more. 3. Journal Rank in Categories This indicator shows the ranking of the journal in its subject categories based on the impact factor. For example, the journal Leukemia is classified into two categories in the JCR. In the hematology category it is ranked number 4 among 62 journals and is among the top 25%. On the other hand, in the oncology category it is ranked number 9 among 141 journals and is also among the top 25%. 4. Eigenfactor Score For the impact factor and the h index all citations are equal with no regard to their origin, while the “eigenfactor is a prestige indicator based on the idea that all citations are not created equal. A citation from a high-prestige journal, such as Science or Nature may count more than one from a local scientific newsletter”. “Operationally, this means that a citation from a highly influential journal will have more “weight” in determining the influence of other journal than a citation from a more modestly successful journal”. “Eigenfactor considers the journal to journal citation network created by citations in the current year to any of the past 5 years. For each journal the Eigenfactor score is a reflection of the percentage of this citation network that is concentrated on this journal”. . Jorge E. Hirsch, “An Index to Quantify an Individual’s Scientific Research Output,” Proceedings. . Tibor Braun, Wolfgang Glänzel, and András Schubert, “A Hirsch-type Index for Journals,” The. of the National Academy of Science 102, no. 46 (2005) : 16569-16572.. Scientist 19, no. 22 (2005) : 8.. http://joemls.tku.edu.tw.

(19) 212. Journal of Educational Media & Library Sciences 49 : 2 (Winter 2011). 5. Article InfluenceTM Article influenceTM is a measure of the per-article influence of the journal. The average article influence of all the papers in the database is set to be 1. The value of article influence for a particular journal being greater than 1 indicates the average influence of papers in that journal is above the average influence of all the papers in the database of JCR. Otherwise, a value smaller than 1 suggests the average one of each paper in that journal is below the average influence. In the JCR, the Article Influence score means “current year citations to 5 prior years from cited journal data tables — scaled by size of the journal’s contribution”. The Journal Analyzer System in the Scopus of the Elsevier also establishes the indexes quantifying the influence of a journal. Among these indexes, the total cited number, total number of papers published for a journal and journal H index are the same as those in the JCR. Other new journal impact indicators proposed and defined in Scopus are listed in the following: 1. Percentage not Cited Percentage not cited means the percentage of articles with zero cited number to the total number articles published in a year. 2. Source Normalized Impact per Paper, SNIP Source Normalized Impact per Paper (SNIP) measures contextual citation impact by normalizing citation values. It takes a research field’s citation frequency into account by weighting citations based on the total number of citations in a subject field. The SNIP “calculates without use of a journal’s subject classification to avoid delimitation and accounts for how well the field is covered by the underlying database. It also “considers how quickly a paper is likely to have an impact in a given field”. 3. SCImago Journal Rank (SJR) SCImago Journal Rank (SJR) is “a prestige metric and based on the transfer of prestige from a journal to another, where the journal prestige is calculated by the number of citations and the prestige of the citing journals. It shares a journal’s prestige equally over the total number of citations in that journal and normalizes for differences in citation behavior between subject fields”. With SJR, the subject field, quality and reputation of the journal has a direct effect on the value of a citation. “It eliminates manipulation and raises the SJR ranking by being published in more reputable journals”. A journal impact indicator can be used as a reference tool to evaluate a journal by citing and cited relationship of the journal. Due to the different sets of  . Scopus, “SNIP,” http://info.scopus.com/journalmetrics/snip.html (accessed January 6, 2012).. http://joemls.tku.edu.tw. Scopus, “SJR,” http://info.scopus.com/journalmetrics/sjr.html (accessed January 6, 2012)..

(20) Ming-Yueh Tsay: An Overview of the Journal Impact Indicators. 213. journals covered, journal impact indicators that reported by the JCR and Scopus for a particular journal are usually different. Though, there are some inevitable disadvantages in the usage, it is still an important tool to investigate the influence of an academic research by the mechanism of citation impact between citing and cited journals. The contents introduced in this paper will hopefully be a useful reference to academic evaluation organization, libraries, journal publishers, database producers, authors and users, etc. ROMANIZED & TRANSLATION NOTES FOR ORIGINAL TEXT. 註 1 H. F. Moed & Th. N. Van Leeuwen, “Improving the Accuracy of Institute for Scientific Information’s Journal Impact Factors,” Journal of the American Society for Information Science 46, no. 6 (1995): 461-467. 註 2 Sture Hansson, “Impact Factor as a Misleading Tool in Evaluation of Medical Journals,” Lancet 346, no. 8979 (1995): 906. 註 3 S. Arunachalam & K. Manorama, “Are Citation-Based Quantitative Techniques Adequate for Measuring Science on the Periphery?” Scientometrics 15, no. 5-6 (1989): 393-408. 註 4 Derek de Solla Price, “Citation Measures of Hard Science, Soft Science, Technology, and Non-Science,” in Communication Among Scientists and Engineers, ed. Carnot E. Nelson & Donald K. Pollock. (Lexington, MA: Heath Lexington, 1970), 9. 註 5 Christinger Tomer, “A Statistical Assessment of Two Measures of Citation: The Impact Factor and the Immediacy Index,” Information Processing and Management 22, no. 3 (1986): 251-258. 註 6 Nevenka Pravdić & R. Pekorari, “The Citing Practices of the Authors to the National Journals in Mathematics, Physics and Chemistry,” Scientometrics 8, no. 3-4 (1985): 233-246. 註 7 S. Arunachalam & K. Manorama, “Are Citation-Based Quantitative Techniques Adequate for Measuring Science on the Periphery?” Scientometrics 15, no. 5-6 (1989): 393-408. 註 8 Amin, M. 演講講義 [Yanjiang Jiangyi] “Journal Publishers and Impact Factors,” in Proceedings of the 8th International Conference on Scientometrics and Informetrics (Sydney,. 16-20 July 2001), 14-15. 註 9 Ibid. 註 10 Leo Egghe & Ronald Rousseau, “Averaging and Globalising Quotients of Informetric and Scientometric Data,” Journal of Information Science 22, no. 3 (1996): 165-170. 註 11 Per O. Seglen, “Causal Relationship between Article Citedness and Journal Impact,” Journal of the American Society for Information Science 45, no. 1(1994): 1-11. 註 12 Eugene Garfield, “The Use of JCR and JPI in measuring short and long term journal impact,” in Council of Scientific Editors Annual Meeting held in 9 May 2000, http://. www.garfield.library.upenn.edu/papers/cseimpactfactor05092000.html (accessed October 29, 2011). http://joemls.tku.edu.tw.

(21) 214. Journal of Educational Media & Library Sciences 49 : 2 (Winter 2011). 註 13 碩睿資訊有限公司 [Shou Ray Information Service],Journal Citation Reports PPT,2010;Elsevier 教育訓練 [Elsevier’s Training],Scopus 資料庫 PPT[Scopus Database PPT],2010。 註 14 J. E. Hirsh, “An Index to Quantify an Individual’s Scientific Research Output,” Proceedings of the National Academy of Sciences of the United States of America 102, no. 46. (November 2005): 16569-16572.. 註 15 F. W. Lancaster,伊利諾大學圖書館與資訊科學研究所 [Graduat School of Library and Information Acience, University of Illinois],書目計量學課程講義 [Shumu jiliangxue kecheng jiangyi],1985 註 16 I. N. Sengupta, “Three New Parameters in Bibliometric Research and Their Application to Rerank Periodicals in the Field of Biochemistry,” Scientometrics 10, no. 5-6 (1986): 235-242. 註 17 Francis Narin & Mark P. Carpente, “National Publications and Citation Comparisons,” Journal of the American Society for Information Science 26, no. 1 (1975): 80-93. 註 18 Bruce C. Bennion & Sunee Karschamroon, “Multivariate Regression Models for Estimating Journal Usefulness in Physics,” Journal of Documentation 40, no. 3 (1984): 217-227. 註 19 Francis Narin & Mark P. Carpente, “National Publications and Citation Comparisons,” Journal of the American Society for Information Science 26, no. 1 (1975): 80-93. 註 20 D. T. Hawkins, “The Percentage Distribution: A Method of Ranking Journals,” Proceedings of the American Society for Information Science 16(1979): 230-235. 註 21 何光國 [James K. K. Ho],文獻計量學導論 [Wenxian Jiliangxue Daolun](台北 市:三民書局 ,1994)[(Taipei: Sanmin, 1994)],187。. http://joemls.tku.edu.tw.

(22)

參考文獻

相關文件

signifies bad financial fortune, the half-peeled a normal year, and the unpeeled a great financial bounty in the year ahead.. To get rid of the financial misfortune of the

Despite significant increase in the price index of air passenger transport (+16.97%), the index of Transport registered a slow down in year-on-year growth from +12.70% in July to

The average Composite CPI for the first half year of 2012 increased by 6.42% year-on- year, of which the price index of Alcoholic Beverages &amp; Tobacco (+29.19%); and Food

Despite higher charges of taxi service starting from September, the index of Transport registered a slow down in year-on-year growth from +8.88% in August to +7.59%, on account of

and Jorgensen, P.l.,(2000), “Fair Valuation of Life Insurance Liabilities: The Impact of Interest Rate Guarantees, Surrender Options, and Bonus Policies”, Insurance: Mathematics

Stone and Anne Zissu, Using Life Extension-Duration and Life Extension-Convexity to Value Senior Life Settlement Contracts, The Journal of Alternative Investments , Vol.11,

The study explore the relation between ownership structure, board characteristics and financial distress by Logistic regression analysis.. Overall, this paper

The difference resulted from the co- existence of two kinds of words in Buddhist scriptures a foreign words in which di- syllabic words are dominant, and most of them are the