Book Profiles and How They’re Made

Syndetics Unbound’s Book Profile is a dynamic tool that lets users browse for titles using clickable profile terms, like character names, genres, topics, locations, and more. There are two kinds of Book Profiles—those created by editorial staff, and machine generated profiles. Today’s blog post is a dive into how Book Profiles are created.

Fiction and Non-Fiction Profiles are curated manually by Syndetics Unbound editorial staff members, who refer to descriptions and reviews found in Industry journals to create a profile for a given title, using standardized terms. When a user clicks on a Book Profile in the catalog, it shows other titles the library has that also fall under that profile term. The below image is an example of a human curated Book Profile.

Academic Book Profiles are a little different. Instead of being created by Syndetics Unbound editorial staff, they are computer generated profiles from the full text of a book or other work using various data mining algorithms. This process extracts several types of attributes from the text of a work, such as a list of the most relevant terms and phrases found in the text. Additional attributes include people, locations, and organizations mentioned in the work, as well as the topics, which are similar to subjects, that the work is about. You can see some of those attributes in the Academic Book Profile below.

All of these attribute values are consistently normalized to facilitate the linking of disparate works that share the same attribute values, as you can see below. By using machine learning algorithms we can analyze a large corpus of academic works to offer patrons an unending list of relevant recommendations to help further their research.