在各類型多媒體搜尋引擎中個人特質對搜尋行為、策略與成效之影響

Abstract

網路資訊搜尋是現代人獲得資訊最主要的方法，因此搜尋引擎設計者在設計產品時應考量使用者的個別差異，以便個體能利用搜尋引擎獲取知識。關於影響搜尋歷程與成效之因素，之前已經累積了一些研究成果，但大多數研究都是探討影響文字搜尋的原因，很少針對其他種類搜尋引擎去探討，但搜尋文字、圖像、影音和地標所經歷的認知歷程是不一樣的，且圖像搜尋、影音搜尋、地標搜尋比文字搜尋更為複雜，故影響其搜尋行為與成效的因素可能更多。本研究要探討個人特質（含認知、情意與行為層面）對各式多媒體搜尋的影響，讓網路搜尋能夠提供更多個人化的服務，幫助使用者提高搜尋的準確率，而不是加重認知負荷。 本研究設計了4階段實驗，分別使用文字（Google）、圖像（Google Image）、影音（YouTube）和地標（Google Earth）搜尋引擎作為研究工具。受試者分別為5~7年級學生，藉由對搜尋歷程的觀察與質性分析以及對搜尋行為與結果的量性分析，進一步瞭解不同特質的受試者在處理各種搜尋任務時會運用哪些搜尋策略與行為，他們的搜尋歷程、成效以及學習成效又是如何。 實驗結果發現在眾多個人特質中「認知方面」的因素影響搜尋行為、策略、成效的效果最為顯著，包括閱讀能力影響圖像搜尋，後設認知能力影響影片搜尋，環境認知能力、空間能力以及地理知識影響地標搜尋。「情意方面」的因素僅有思考風格（幅度）會影響文字搜尋，認知風格則對影片搜尋沒有影響。「技能方面」的因素，網路使用經驗對圖像搜尋沒有影響。因此學生認知的能力在網路資訊檢索中扮演重要的角色。此外，我們發現在不同形式（文字、圖像、影片、地標）的搜尋引擎中，使用者必須使用不同的搜尋策略（例如：在文字搜尋中要使用多個關鍵字並靈活變化，但在圖像搜尋中要多比對審視搜尋結果，而在影音搜尋中則要鍵入精簡明確的關鍵字，並根據目標簡要審視搜尋結果），才能尋獲正確且符合需求的資料。

Web information retrieval is, for most people, the most popular method of acquiring information. Therefore, when creating search engines, designers should take into account human factors (including cognitive, affective, skill, and demographic factors) in order to increase the degree of user-friendliness. Many researchers have examined the factors impacting information-seeking behaviors, strategies and performance and have accumulated fruitful results, but they have generally focused on text rather than on other multimedia searches. However, the cognitive processes underlying search for texts, images, videos, and landmarks are very different. Search for images, videos, and landmarks require more complicated procedures than search for texts do; therefore, more factors could influence search behaviors and performance in these complex processes. This study aimed at investigating how individual factors influenced a variety of multimedia searches (including search for texts, images, videos, and landmarks). We hope that search engines can provide more personalized services and that these personalized functions will push for the accuracy of search results instead of pulling for greater burdens on cognitive loads. A four-phase study was conducted to explore multimedia searches with increasing complexity. The search engines “Google” was used in search for texts, “Google Image” was used in search for images, “YouTube” was used in search for videos, and “Google Earth” was used in search for landmarks. Study participants were fifth through seventh graders. This study was conducted to explore the impacts of (a) thinking style level on search for texts; (b) reading ability and Internet experience on search for images; (c) metacognitive strategies and verbal-imagery cognitive style on search for videos; and (d) geographical knowledge, spatial ability and environmental cognition on search for landmarks. As a result of the quantitative data analysis and qualitative process observations, I can explore the participants with different characteristics used what types of search strategies and behaviors when they undertook various search tasks. Furthermore, I can understand the search performance and learning effectiveness of the participants. The results support the conclusions that many cognitive factors significantly influenced search behaviors, strategies and performance. Examples include “reading ability” influencing image search; metacognitive strategies influencing video search; and “environmental cognition”, “spatial ability”, and “geographical knowledge” influencing landmark search. In terms of affective factors, only thinking style levels were significant factors influencing text search, whereas verbal-imagery cognitive style did not exert a strong influence on video search. A skill factor (Internet experience) did not influence image search. In sum, cognitive abilities, compared with affective, skill and background factors, played more important roles in web information retrieval. Moreover, our results show that the users adopted different search strategies when using the various types of search engines, such as using more keywords in text searches, evaluating more search results in image searches, and entering concise and accurate descriptions with fewer keywords in video searches.