Do you sometimes want to get right to the juicy bits? Are you ready for action? Or is it all about the bottom line?

Text Textures makes page-by-page profiles of what's going on in a book. This can be used by readers, publishers, booksellers, and other stakeholders for diverse applications.

Readers can display "page-lines" (like time-lines) to anticipate where particularly sexy, violent, or affectionate scenes occur. Or for a different kind of book, the "page-lines" could help you navigate to where money matters are discussed. Or religion, or the occult. You can also look for combinations, like fights about money.

You can also display text textures in a different mode, to get global impressions. Is this a book with lots of sex and violence, more of one than the other, or not much of either? Readers can build profiles from different perspectives, to compare what they're reading to other books -- books they've liked or are thinking of trying.

We've also combined the information from many categories to construct mathematical profiles of the overall "texture" of books. These profiles are at work under the hood of a recommendation engine. The engine describes the texture of a chosen book, directs readers to the most compelling passages, and recommends a selection of similar books. You could even just indicate in a general way what kinds of content interest you, and get back appropriate recommendations.

Text Textures should interest not only readers, but also authors, booksellers, and publishers. For example, using Text Textures, we can generate a ranked list of similar titles from the raw text of an unpublished work. This could help in matchmaking with appropriate agents, early reviewers, and opinion-makers prepublication.

Text Textures is at the prototype stage today, but all its components are scalable. The possibilities for features built around the underlying technology are open-ended. For example we've played with the possibility, in an e-reader application, of automatically generating appropriate mood music or background images based on what's happening around where you're reading.

Share this project: