評估以MeSH做為PubMed資料庫搜尋之建議詞彙的有效性檢索行為研究 A User Study of the Effectiveness of MAP (Multiple-Access to PubMed): A MeSH Based Query Suggestion Tool for PubMed

本研究設計了一針對PubMed資料庫的詞彙建議系統介面,並以實驗法比較該介面與PubMed介面。以44名生物科學及醫學系所研究生為研究對象,每人以兩個自己的真實需求作為檢索任務,就使用者查詢行為及檢索結果進行分析。研究結果顯示:(一)主題熟悉度會影響使用者檢索詞的修改,不熟悉主題任務其檢索前後相似度會低於熟悉主題任務。(二)不同介面會影響使用者查詢行為,實驗介面能夠提供使用者大量的建議相關詞彙,因此使用者改變檢索詞的機率增加,而在PubMed介面中使用者查詢詞彙的變化有限。(三)不同主題熟悉度與不同介面對使用者查詢行為產生之交互作用,實驗介面在使用者較不熟悉檢索主題時可以發揮較大的作用。(四)實驗介面能夠輔助使用者取得在大量檢索結果中排序較低的相關文件。 An interface, MAP (Multiple-Access to PubMed) was proposed here to provide MeSH term suggestion for PubMed bibliographic search. This paper reported the results of an experiment comparing users’ search behaviors and performance between MAP and regular PubMed interfaces. A total of 44 graduate students in Bio-medical domains in Taiwan participated in the study. They were asked to ask for two information problems of their own alternately with the two interfaces. Comparison were made between the interfaces in terms query “goodness”, result quality, and impact on users’ querying behaviors. Information regarding the attributes of their information problems was also collected through questionnaire so that the effectiveness of the interfaces with different ty0pes of information problems can be explored. The results show that 1. Topic familiarity would influence users’ querying behaviors; MAP had a high impact on users’ queries when searching for unfamiliar problems. 2. MAP was shown to be helpful, especially when searching for search requests that the users had not attempted before. 3. MAP was most helpful in generating new research ideas. 4. MAP was shown to be able retrieve documents buried deep down in the initial returned results.