小說讀者資訊尋找心智模型與FRBR之比較研究

本研究旨在探討小說讀者資訊尋找心智模型，利用概念圖法與訪談法引導小說讀者尋找資訊的方式，再以資源描述框架模式標誌方式表徵讀者繪製的心智模型。此外，對照讀者心智模型至書目記錄功能需求（Functional Requirements for Bibliographic Records, FRBR）與權威資料功能需求（Functional Requirements for Authority Data，FRAD）模型的實體與關係，比較兩者異同情形，期能提供小說資訊描述相關建議，並運用於小說資訊組織實務。本研究共歸納為41種資料元素並分為九大類外，亦發現元素出現次數並未具備長尾定律、80/20法則與冪次定律等現象。讀者以多元的基本路徑尋找小說，以出現頻率高的書名與作者等元素作為起點。書名作為終點次數極端，代表其他元素較少輔助尋找外的使用者任務，中介元素則具有齊夫定律的最小努力原則現象，且可經由二至四個元素達成需求。作品類、出版類、作者類、外觀尺寸類與故事內容類元素導航功能較高，但讀者已知資訊可能影響作者類的導航。心智模型對照至FRBR與FRAD模型方面，基本路徑分為完全對照、部份可對照與完全無法對照三種。主要研究結果如下： 1. 39.8%路徑可完全對照至FRBR與FRAD模型實體關係。讀者主要需求為作品與內容版本間摘要、補充、續集、實現與插圖關係，重視第一組與第二組實體的創作關係，並以多樣的團體與載體版本關係瞭解小說。路徑可能因FRBR與FRAD模型部分實體與關係定義較不明確而對照至複數關係。對照複數關係的路徑中有9.1%實體關係屬於權威資料功能需求，且提供32.5%導航路徑，可能代表FRBR與FRAD模型導航功能不如預期。 2. 部份對照路徑數量佔總體34.8%，也常被小說讀者使用。讀者易建立作品與單件的直接路徑。常被提出的個人與團體協同關係無相符實體關係。讀者多用第三組實體表示故事內容，且不視為主題。讀者鏈結個人與第三組實體以瞭解作者相關資訊，第三組實體間的鏈結用於表達故事背景。前述鏈結與內容版本摘要屬性資訊類似，推測是因資訊粒度判斷差異，或摘要屬性範疇不明確。 3. 時間與感受元素相關路徑完全無法對照，佔總路徑數25.3%。時間在讀者未知意義時需求相對提高，主要滿足FRBR尋找任務需求。感受元素顯示讀者喜歡以主觀判斷尋找小說資訊，依鏈結對象等結果，感受元素應被用於辨識任務。感受元素是讀者主觀評價判斷，不宜視為內容版本評論屬性。本研究建議FRBR與FRAD模型明確界定第一組書目實體範圍，並具體定義特定實體關係。此外，應再研議實體屬性種類及使用方式，根據既有實體關係開放更多語意關係。最後，建議FRBR與FRAD模型給予時間與感受相關元素相應位置。

關鍵字

心智模型；鏈結資料；書目記錄功能需求（FRBR）；權威資料功能需求（FRAD）；小說資訊描述；使用者任務

並列摘要

This study is aim to explore fiction readers’ mental model of information finding. Interview and concept mapping method are used to guide participant to show their mental models as linked data by Resource Description Framework (RDF) triple. This study also mapping participants’ mental model to entities and relationships defined by Functional Requirements for Bibliographic Records (FRBR) and Functional Requirements for Authority Data (FRAD). This research expect to recommend for practice of fiction information description, by find out differences between fiction readers’ mental model and FRBR and FRAD. Research result in section of fiction reader’s mental model, consolidate 41 kinds of data elements from participant concept maps, and classified in nine categories: Works, Author, Story Content, Physical Dimension, Publishing, Feeling, Reviews, and Experience. Data elements occurrence frequency do not show the phenomenon of The Long Tail, 20/80 Law and Power Law. Fiction readers use multiple routes to find fiction they need, and commence the route with elements with occurrence in high frequency, like title of work and author elements. Title of work also has high frequency to be the end of routes, this represents that other data elements were less to use for other user tasks defined by FRBR, beside find tasks. Intermediary elements’ occurrence frequency conformed principle of least effort in Zipf’s law. Two to four elements can meet fiction reader’s demand of finding. Elements from class of Works, Publishing, Author, Physical Dimension, and Story Content, provide higher functionality of navigation to reader. Author class elements' navigation may influenced by reader’s known information. Research results in section of mapping reader’s mental model’s routes to FRBR and FRAD entity relationships are divide into three parts: complete mapping, part mapping, and unable mapping. Main finding of each part are as follows: (1) Totally 39.8% routes complete mapping to FRBR or FRAD relationships. Work/Expression entity relationships of summary, complement, successor, realization, and illustrate of text, are mostly needed by fiction readers. Fiction readers attach importance to FRBR Group1/Group2 relationship of Create, and use multiple routes of Manifestation/Corporate Body to aware the fiction they want. Route may mapping to multiple relationship when FRBR and FRAD defined entity relationships unclear. Relationships that FRAD relationships rate 9.1% of all multiple mapped relationships, and the relationships provide navigation routes in 32.5%, imply navigation function of FRBR and FRAD possibly not as expected. (2) 34.8% routes can part mapping FRBR or FRAD relationships. Routes in this part link most number of different data elements, and has higher utility rate. Fiction readers prefer to link entities of Work and Item. The Corporate Body/Person relationships, that fiction readers frequently proposed and consider as coordination relationship, do not exist in FRBR or FRAD. Fiction readers use FRBR Group3 entities to display story content, rather than as subject of work. Fiction readers use routes between FRBR Group2/Group3 to understand authors’ information, and use routes between FRBR Group3 four entities to understand story content. Routes that between Group2/Group3 and Group3 four entities resemble the summary property of Expression, difference of granularity of information that participants identified and unclear definition of Expression summary property are inferenced as causes of this phenomenon. (3) Routes related to feeling relate elements and Time-span elements take up 25.3% of all, unable mapping to relationships of FRBR and FRAD. Time-span elements’ has higher importance when the meaning of time-span not realized by fiction readers. Feeling relate elements show that fiction readers prefer to find fiction by self-judgment. Based on results like the kinds of elements linked to feeling relate elements, feeling relate elements possibly be used in FRBR user task identify. Feeling relate elements origin from readers’ subjective judgment, it possibly different to critical response property of Expression. Based on research finding, Suggestion for FRBR and FRAD are as follows: (1) FRBR and FRAD supposed to clear defined the scope of FRBR Group 1 entities, and make relationships definition more specific. (2) Entities may need to discuss the properties and usage, defined more semantic relationships that based on existing relationships. (3) FRBR and FRAD supposed to accede to feeling relate and Time-span elements.