• English
Innovation and improvement

How Kindred Works … works

A look inside the new OCLC Research demo for content-based recommendations

There are many ways to find a new book to read or movie to view. For example, you can ask a librarian, consult websites that provide suggestions based on consumer purchasing behavior, or rely on the recommendations of other users. OCLC Research has developed an experimental service that provides a list of items similar to an item of interest. The prototype service uses various characteristics of a sample work, such as classification numbers, subject headings and genre terms, to retrieve related resources from WorldCat. This approach is called content-based recommendation.

The recommendations are accessible through a user interface and through an application programming interface (API). The user interface, Kindred Works, provides basic search functionality; users can search by author, title, ISBN or OCLC number. If more than one record is returned, the user is presented with a list of items to choose from. The recommendations may include books, e-books, audiobooks, music and video materials. The WorldCat Search API is used to retrieve results.

The Kindred Works interface provides a convenient means for viewing the results of the WorldCat Recommender API. The Recommender API requires an OCLC number or an ISBN to operate. A high-level architecture diagram of the Kindred Works interface and Recommender API is shown in the diagram to the right. Once the Recommender API is supplied with an OCLC number or ISBN, a supplemental database is searched to obtain work-level characteristics associated with the resource. These characteristics include:

  • Dewey class number
  • Library of Congress class number
  • Fiction indicators
  • Genre terms
  • Audience indicators
  • Subject headings

A query is generated based on these characteristics and sent to the WorldCat Search API. The results of the query constitute a set of related items. The recommendations can be customized to the collection of an individual library by adding the library’s OCLC holding symbol to the query. This feature is limited to libraries that participate in The machine service is intended to be used by software developers to integrate recommendations into another service or application, for example, a library catalog or other discovery interface. As an experimental service, the Recommender API is likely to be extended and refined over time. Similarly, the Kindred Works interface may evolve based on user feedback.


The user interface is available at

The API is available at Technical information about the Recommender API is accessible from the Kindred Works home page and through the OCLC Developer Network.