Identifying, evaluating and recommending digital collections: A social community approach
We are developing a community based digital collection repository. The goal of the system is to increase the visibility and use of digital collections by forming a publicly accessible online social community where everyone can share their knowledge and experience of using different collections. In this poster, we present our design of a digital collection system that will respond to the challenges of identification and organization of digital collections as well as incorporate a process for reviewing, ranking and recommending digital collections through the creation of an online digital collection community.
Although there is an abundance of digital collections offering content of cultural interest, educational value, and historical significance, it is still difficult for people to locate collections which satisfy their information needs. A failure to agree upon a universal metadata standard for describing digital material  makes retrieval of collections with web search engines difficult. To counter this, several tools have been developed which aim to help users locate digital materials. OAIster  is perhaps the largest and best known of these, but since it harvests metadata at the item level, finding entire collections using OAIster can be frustrating. Furthermore, there is as yet no mechanism within OAIster for browsing by subject. Some of OAIster's organizational difficulties are addressed by the Digital Collections and Content  and Opening History  tools developed by the University of Illinois. These registries offer the ability to browse collections at the subject level as well as search for individual items. Despite the developments of such tools there is still much to be done to maximize the usage of digital materials. Our digital collection repository system will enlist the user community to assist with the identification, evaluation, and organization of digital collections, and provide mechanisms for submitting and obtaining collection reviews, rankings and recommendations.
COLLECTION IDENTIFICATION & DISCOVERY
Digital collections of high quality have been developed by public and private institutions alike, and the Institute of Museum and Library Services (IMLS) continues to support digital collection creation and development. Even so, digital collections cannot achieve their full potential while they remain difficult to find. Each digital material finding tool develops its own criteria for inclusion: for OAIster it is the presence of OAI-PMH, for Digital Collections and Content it is collections which have been supported by IMLS funding. After exploring the kinds of materials included and excluded by other tools, our first step in designing a repository that is inclusive of the best publicly available digital collections was to identify a set of core collections based on the digital collection output of the nation's top universities and public libraries. This core group of collections will be enriched by the recommendations of librarians, library school students, and the general web community, as well as through the implementation of automatic harvesting techniques.
Identification is only a first step; the next step is organization. Presently, OAIster does not offer searching or browsing at the subject level, something we feel impedes the usefulness of the system. The Digital Collections and Content and Opening History repositories organize collections in part by facilitating browsing using GEM (Gateway to Educational Materials) subject headings. This classification schema offers a useful and relatively uncomplicated way of describing online resources, but it was too broad and educationally focused for our needs. Instead, we are working on a bottom-up organizational scheme based on our core collections that is descriptive, dynamic, and intuitive and that can grow as the collections grow. We also hope to use folksonomies to support and enhance more traditional and structured forms of classification.
REVIEWING, RANKING, & RECOMMENDING
Current digital collections finding tools do not have a systematic method for evaluating their collections nor do they offer users the opportunity to review, rank or recommend collections after usage. Users are unable to determine the value of one collection in relation to other similar collections or indicate their own views of a collection. Our design incorporates a systematic online evaluation and review process based on a user-centered framework developed from a review of the literature. This was supplemented by data gathered from a Digital Libraries class about the types of evaluation criteria judged to be important, and is presented in Table 1.
Table 1. Types of Evaluation Criteria
Based on a user-centered framework, evaluation will be done using an easily accessible online form that incorporates the evaluation criteria with specific parameters for a consistent evaluation of digital collections. The systemization of the evaluation will give users the tools to compare digital collections and to assess the value of a collection. A separate online review form will allow user to submit their own reviews based on their personal criteria and usage of the collection. It is a quick and effective method for users to share their views about a collection with the community. In addition, users can judge digital collections on a scale of 1 (low) to 10 (high) during the evaluating and reviewing processes. Our design creates a robust digital collection system based on hybrid recommender system technology , and our algorithm will rank digital collections based on explicit reviews and ratings submitted by users, and generate automatic system recommendations based on user and collection data. With a click of a button, users can submit member recommendations to other members of the community. Thus, users will receive both automatic (system-generated) and member (user-generated) recommendations from our digital collection system. We will also include Web 2.0 technologies that will allow users to search all content associated with a collection (descriptions, reviews, rankings, recommendations), create their own personal collections, and develop groups/networks. Similar to WikiLens , recommendations are community-based and the system exists as a tool that enhances what the community already does naturally.
Many librarians already identify, evaluate and recommend digital materials for their patrons, but there is yet to be an online system that does all three at once. Digital collections are still difficult for users to locate, and once found an individual needs to invest a significant amount of time to assessing their usefulness. Our system focuses on identifying, organizing, reviewing, ranking, and recommending digital collections, and will remain dynamic and relevant to the interest of our users while building a community around digital collections. Our hope is that it will become a tool that librarians and anyone developing digital collections will contribute to as well as share with others for mutual benefit.