Volume 4 Issue 1
Spring 2008
ISSN 1937-7266

ContextMiner: A Tool for Digital Library Curators

Chirag Shah

School of Information & Library Science
University of North Carolina
Chapel Hill NC 27599, USA

Gary Marchionini

School of Information & Library Science
University of North Carolina
Chapel Hill NC 27599, USA

The VidArch Project aims to develop policies and tools that help video curators discover and add contextual elements that will help future generations not only find but also make sense of video content.

In order to help the curator in fetching and evaluating information for its relevance, we have developed a prototype system called ContextMiner. This web-based tool, implemented primarily using PHP and MySQL, helps the curator in looking for information on the Web or in specialized databases, compiling it, and finally storing it in the repository. The primary interface of ContextMiner consists of an integrated set of tools that are designed for one single task – provide the curator as much information as possible relating to the object she is assessing for the collection.

The interface in its initial condition is presented in Figure 1. The screen is divided in two panels. The panel on the right consists of a form with a number of fields that represent certain metadata or contextual information about the object being collected. In a full system, these fields would be mapped to the local content management system.

The panel on the left has four tabs: Search, OpenVideo, Prelinger, and YouTube. One can toggle between these tab without reloading the whole page. All the tabs, other than the ‘Search’ tab, contain a form similar to the one displayed on the right panel. The ‘Search’ tab includes a search box where the curator can enter a query. Along with this query, the curator can specify which source to execute this query in. This can be indicated by selecting the source from the dropdown box with three options: OpenVideo, Prelinger, and YouTube. A sample result list for query ‘space telescope’ in OpenVideo is shown in Figure 2 (note only the result panel is show to give more readable detail). When the curator clicks on a result, the system fetches as much information about that object as possible from the given source. At this point the left side panel automatically switches to the corresponding source tab and fills in the fetched information in various fields on that tab. Figure 3 shows the result of clicking on a search result in the ‘Search’ tab that came from YouTube. As we can see, the left side panel has most of the fields of the right side. Each field on the left is followed by an arrow button. Clicking on this button transfers the content of that field to the corresponding field on the right. Thus, if the curator is satisfied with the automatically extracted information, she can collect them in just a few clicks.

In summary, the interface on one side provides the curator a workspace, where she can search on and extract information from different sources, compare and compile them, and transfer them to the form on the other side to store in the collection.

Figure 1: The curator interface (click here for a larger view )


Figure 2: Searching in OpenVideo


Figure 3: Result of clicking on a YouTube search result. The form on the right shows some of the fields filled in using the metadata automatically extracted on the left.
(click here for a larger view)