資源合集層次描述應用於臺灣數位典藏之研究

Abstract

數位典藏與數位學習國家型科技計畫將國家級的重要典藏文物數位化，累積已產出超過300萬筆詮釋資料記錄，但隨著資料量日趨龐大，使用者變得難以迅速檢索與使用這些資訊，管理者亦難以確實掌握所擁有的整體資源。為解決此困擾，本研究試圖建立臺灣數位典藏資源合集層次描述雛型，以為未來建立數位典藏資源合集層次描述之基礎，並將此建置經驗提供國內作為參考。 本研究以國立故宮博物院、國立自然科學博物館、國立臺灣大學與國史館此四機構下所產出之數位化資源為研究範圍，使用都柏林核心集資源合集應用綱要（Dublin Core Collection Application Profile, DCCAP）作為描述標準。訪談34位研究人員與數位典藏計畫執行人員，整理出73個資源合集，依其情況與藏品特性，修改DCCAP描述標準，並建立資源合集層次描述。最後利用層面分類的方式呈現於資訊系統中，改善使用者進行資源的瀏覽和利用。經分析整理後，本研究有以下幾項發現： 一、 資源合集層次描述的建立可歸納為五個階段。首先須依據目的，確定資源合集範圍集描述深度；再來按照目的、標準聲譽、採用計畫、相容性、簡易性、軟體工具、與維護情形七種面向選擇合適的描述標準；接著利用訪談等方式瞭解資源合集內涵，建立資源合集層次詮釋資料；設計呈現介面，將產出成果加以組織呈現，改善資源的瀏覽使用；最後進行系統評估，瞭解資源合集層次描述之成效。 二、 數位典藏資源合集層次描述具資源合集異質性極高、資源合集層次描述不斷變動、資源合集層次描述的建立需要高度的學科知識、機構角色也會影響描述品質、文化類資源合集層次描述之品質優於自然類資源合集等現象。 最後建議由數位典藏執行單位推動完成數位典藏資源合集層次描述之建立。最好由兼具學科背景與資訊組織知識者撰寫詮釋資料，以物性為主進行描述，並維持單件藏品的獨特性與重要性。可依目的修訂描述標準，以平實用語取代專業術語，增近使用者理解。同時建議善用層面分類組合功能，建立資源合集與單件的連結，改良系統功能。未來可進行使用者研究，評估運作成效，或進一步探討自然與人文領域對資源合集層次詮釋資料適用性之異同。

Taiwan e-Learning and Digital Archives Program (TELDAP) is dedicated to digitizing Taiwan’s cultural heritages. So far, TELDAP had already comprised of a remarkable digital collection of over 3,000,000 metadata. With the ever increasing amount of data, users are starting to encounter difficulties while searching for and using information contained. Administrators also are starting to find it difficult to truly manage the overwhelming amount of data. Aiming at solving such problems, this study tries to build a collection-level-description (CLD) for Taiwan’s Digital Archives. It is hoped that this study can become the foundation for future Digital Archives Collection Level Description building and serve as a reference for other databases. The scope of the study covers digital archives produced by National Palace Museum, National Museum of Natural Science, National Taiwan University and Academia Historica. 34 researchers and program implementation staffs are interviewed to identify 73 collections using Dublin Core Collection Application Profile (DCCAP), as the CLD schema. According to specific conditions and special features of the collections, the DCCAP is modified to create an applicable CLD. Lastly, faceted classification is adopted by the information system for CLD presentation to improve the users’ experiences in browsing and using of the resources. With detailed analysis, findings of the study are as follows: 1. CLD’s creation is a five-stage process. The profundity of collections’ scope description is first identified. Then, an appropriate CLD schema is selected according to its fitness for purpose, reputation, existing experience, compatibility, easiness, tools used and sustainability. Interviews are later held to obtain a better understanding of the collections and create a collection-level metadata. The user interface is designed after the above stage to present the outcomes in a systematic fashion for better resource browsing and using. Finally, a system evaluation is conducted to identify CLD’s performance. 2. Digital archives CLD’s characteristics include: high heterogeneity of collections, CLD changes constantly, high level of academic knowledge needed for CLD creation, description quality is influenced by resource producing institutions, CLD for cultural artifacts enjoys a better quality than for science-themed collections, etc. The study suggests that the creation of TELDAP CLD is carried out by the TELDAP execution organization. Metadata should be produced by information bearing personnel with an academic background from the knowledge organization. Collections are described according to their property, and special attention is paid to detailing the unique and important qualities of items in the collection. CLD schema can be modified according to its purpose. For example, the use of technical terms is minimized to enhance users’ understanding of the information. Faceted classification is another point of focus that the study proposes. It can create a relation between collections and individual items while at the same time improve the system’s function. In the future, user experience research shall be conducted to evaluate the operational performance and further explore the differences between its applicability on the collection-level metadata of nature and cultural domains.