Yewno: Transforming Data into Information, Transforming Information into Knowledge

Abstract

Libraries have long been concerned with the transition to the Semantic Web. Studies have shown that many patrons begin their search for library materials on the Web as opposed to a local discovery layer, making the representation of library metadata on the Semantic Web essential for a library’s survival. Most often this is accomplished through the transformation and serialization of a library’s bibliographical metadata encoded in Machine Readable Cataloguing, or MARC, to linked data in a commonly used ontology such as schema.org or BIBFRAME. But this linked data is derived from expensive, handcrafted bibliographic surrogates for a shrinking percentage of resources in a library’s collection. As full-text digital resources and datasets begin to dominate library collections, this handcrafted approach to bibliographic metadata creation cannot scale. Yewno, through such offerings as Unearth, complements traditional library discovery by providing structured access to these burgeoning collections through the use of Artificial Intelligence.

Yewno begins by extracting entities such as names, places, and dates from digital text. In addition, Yewno extracts concepts from the entire data store and relates them to each other in large graphical structures. Numerous features such as journey mapping, knowledge map layering, and concept expansion allow the user to explore the unstructured data making use of concepts extracted directly from the data. The presentation will include a live demo of Yewno’s capabilities (with canned back-up in case of poor connectivity).

The world of discovery and access is in a time of impatient transformation. Through the conversion of metadata surrogates (cataloguing) to linked data, libraries can represent their traditional holdings on the Web. But in order to provide some form of controlled access to unstructured data, libraries must reach beyond traditional cataloguing techniques to new tools such as artificial intelligence to provide consistent access to a growing world of full-text resources.