網路問答服務之回答引用來源分析研究

A Study on the Cited Sources of Answers on Web Question Answering Service

何怡融(Yi-Rong He)

指導教授 ： 卜小蝶

國立臺灣師範大學/教育學院/圖書資訊學研究所/碩士(2011年)

摘要

網際網路的普及讓資訊的傳遞更為便利，上網查找資料已相當普通，然而關鍵字搜尋到的資訊數量往往過於龐大，而網路問答服務提供「以問代查」的方式，讓使用者節省過濾資訊的時間，甚至集結眾人智慧解決問題，輕鬆又有效率的取得資訊。本研究主要目的在分析網路問答服務平台「Yahoo!奇摩知識+」，回答內容所引用的資訊來源，引用來源的特性包括：引用來源管道、類型、主題分類。此外，再藉由可及性、重覆性及可靠性等，判斷引用來源的品質。
本研究採內容分析法，並以網路問答服務平台「Yahoo!奇摩知識+」為研究對象，抽樣期間為2010年3月1日至2010年3月20日，扣除「煩惱心事」類後的十大類別，每類的前1,000筆問題，共計10,000筆問題進行分析。總回答數為23,874筆，平均每則問題有2.39筆回答；總引用來源數為16,239筆，平均來源數為0.68筆。引用來源的主題分類主要以知識+的十大類別為依據，再以當中重要的問題主題類別進一步區分為十七類。
研究結果顯示，「人際引用來源」（49.56%）及「網路引用來源」（49.05%）的可及性高，為最重要的引用來源管道；「自身經驗」是人際引用來源中最主要引用的來源類型，在網路引用來源中，仍以Web1.0網頁所提供的訊息居多，在Web2.0網頁中，「部落格」為主要引用的類型，而網域分類上以「.com」引用最多；在引用來源的主題分類上，「生活資訊」所佔的比例最高，而在知識+類別中「商業金融」類的引用來源重覆率最高，其回答內容的可靠性較為不足。此外，這些引用來源具有以下特性：發問內容豐富，引用來源的管道多元；提出問題者以初階使用者為主，回答者則以高階使用者較多，在回答內容中標示引用來源有助於讓回答更為完整，有助於判斷最佳解答。最後本研究建議，網路問答服務之使用者應可嘗試客觀評估資訊、並多方求證，而網路問答服務之提供者，也可提供資訊引用來源相關規範，將有助提升網路問答的品質。

關鍵字

網路問答 ； 網路問答服務 ； 引用來源

並列摘要

The widespread use of internet has made the share of information and communications more convenient. It is common for users to surf the web for information. The keyword search engine contains a broad range of information. However, the internet questioning and answering services provide a way to search for information by inserting a question. This type of service allows users to save time filtering information and so they are able to collect information more efficiently. The purpose of study was to analyze web question answering services "Taiwan's Yahoo! Knowledge+", replied the content of citing sources, include: cited source channel, cited source type and cited source theme category. In addition, and then by accessibility, repeatability and trustworthiness, to judge the quality of the source cited.
This study uses Content Analysis. We collected data using a crawler that browses Taiwan's Yahoo! Knowledge which discusses knowledge-based questions. The sampling period is from March 1, 2010 to March 20, 2010. From this pool of data, we net of "Anxieties and feelings" after class ten categories, each type of data before the 1000, for a total of 10,000 data. The total number of answers is 23,874; the total number of cited sources is 16,239. Thus, the average number of answers per question is 2.39; the average number of cited sources per answers is 0.68. Lastly, we randomly selected a total of 5,391 sources for further analysis. Cited source theme category based on Yahoo! Knowledge+ of Categories was then divided into major theme categories 17.
Our results show that the most frequently cited sources are Human cited source (49.56%) and Internet cited source (49.05%). According to the source distribution by genre, “personal experience” was main genre in human cited source; “blog” as main genre in internet cited source. The majority of the Internet Cited Sources are from the ".com" domain. The subject category of “Life information” is the most cited subject category. “Business & Finance” has the highest rate of repeatability of cited source classification. Since certain “Business & Finance” web materials may sometimes be unreliable. Therefore, determining the cited sources of will help us analyze the quality of the answers.
In addition, cited source has the following characteristics: (1) Rich to ask questions, citing multiple sources of channel. (2) In most cases, primary users to ask questions and senior users to responses answer. Cited sources in response to the content can help to answer more completely, will help determine the best answer.Finally, we recommend Web question answering services users can try to assessment objective and confirmation of information.Web question answering services provider can provide information related specifications cited sources, will help to enhance the quality of the Web Social Q&A.

並列關鍵字

Web Social Q&A ； Web question answering services ； Cited source

參考文獻

黃慕萱（民83）。情境模式在解釋資訊尋求行為之適用性探討。圖書館學刊，9，49-70。

卜小蝶、黃斐籃（民96）。網路問答服務系統探析：以台北市立圖書館線上參考服務系統與Yahoo!奇摩知識+為例。圖書與資訊學刊，62，35-51。

Alexander, J. E., & Tate, M. A. (1999). Web wisdom: How to evaluate and create information quality on the web. Mahwah, N.J. : Lawrence Erlbaum Associates.

Berelson, B. (1952). Content Analysis in Communication Research. Glencoe, Ill.: The Free Press.

Bower, H, 1996, Internet sees growth of unverified health claims: BMJ, 313(7054), 381.