What is Talpa Search?
Talpa Search is a new way to search for books and other media. Talpa combines cutting-edge technology with data from libraries, publishers and readers to enable entirely new ways of searching—and find what you're looking for.
For Libraries and Others
Talpa was designed for libraries, to enhance their existing catalog search. Our library version is available by subscription to libraries, has no usage caps, and is integrated with the library catalog and the library's holdings.
If your libary is interested in Talpa, contact [email protected] for more information, for a quote, or a free trial.
An open, non-library Talpa is available to users who sign in with a Google account. Non-library users are currently restricted to 50 searches per day. This version of Talpa is not linked to any library holdings or library catalogs.
Talpa is also available as an API, to be implemented by booksellers and other book-world companies. Contact [email protected] for more information.
How to Use Talpa
Talpa finds books and other media in two ways:
- Search Books. Use natural language to find items by plot details, genre, descriptions, and much more.
- Search Covers. Search for items by the visual components of the cover, including the color and what's on it.
Book Search
Book Search allows you to use plain English to search for books. Many different types of searches work.
Click to see examples of these different types of searches.
Search Covers
Search Covers can find books and other media from the cover color, cover description and the genre or subject.
Some examples:
Frequently Asked Questions
How Does Talpa Work?
Talpa combines traditional library and book-industry data with new artificial-intelligence ("AI") technologies. If you're interested in the technical details, we've written a longer section about this.
Is Talpa always right?
No. Talpa is a search engine, not a truth engine. Its goal is to find material that may be relevant to your search, but it nearly always returns results and the results may include both great and less-than-great answers.
How do I report a problem?
If you see inappropriate or incorrect search results, you can give us feedback by using the comment form at the bottom of this page. You can also reach out to us at [email protected].
What do the bullseyes mean?
A bullseye marks search results that Talpa is particularly sure about. For example, if you search for "Mark Twain book about Huckleberry Finn," the bullseye highlights the answer it thinks is a really strong match. If your search is more general, you won't see the bullseye.
Can Talpa find things other than books?
Yes! You can use Talpa to search for movies and music. You can also mention what you're looking for in the query, such as "1980s teen movies" or "audiobooks about financial scandals."
Does Talpa train AI on my data?
No. Talpa does not train AI systems on your data.
Where does Talpa get its information?
Talpa Search gets its data about books and other media from several main sources:
- Libraries. Talpa draws on over 70 million library records from around the world. When used by a library, Talpa is integrated with the library's holdings and online catalog.
- Publishers. Talpa draws on data from Bowker Book Data and Syndetics Unbound, developed by ProQuest and LibraryThing.
- LibraryThing. Talpa's recommendations and some of its other data, such as series and award information, comes from LibraryThing.com and its community of over three million members.
Where does your cover data come from?
You may imagine that the image labeling—that is, knowing what's on the covers—is done by AI, but actually much of it was done by humans playing a game. Since 2010, thousands of members of LibraryThing have been playing "CoverGuess," a game where members vie to describe covers, getting points when their descriptions match those of other players!
Who developed Talpa?
Talpa was developed by developers, librarians, and librarian-developers of LibraryThing. Find out about us here.
What does Talpa mean?
Talpa is the Latin word for "mole." Moles love to dig and find things! Also, our main developer is a big Latin fan.
How Does Talpa Work?
If you're interested, here's some more information on how Talpa works.
Talpa uses "artificial intelligence" (AI) technologies in two basic ways.
First, Talpa queries standard and customized Large Language Models (LLMs) for answers. LLMs can "hallucinate" books when they don't have good answers. Talpa overcomes this by checking all its answers against a database of over 30 million titles, and 100 million editions.
Second and more creatively, Talpa uses Large Language Models to interpret or "parse" ordinary language into more traditional searches. For example, if a user searches for "fiction about france during world war II," Talpa parses this into three searches—for "fiction," "france" and "world war II"—and then combines the results. It also finds books by translating searches into library and book-industry classifications and subjects, such as the subject "World War, 1939-1945 -- Fiction -- France." This method is particularly useful when it comes to searches about series, awards, publication dates, media, language, settings, and editions. Talpa has exact and complete information on these topics—far beyond what LLMs know.
The combination of techniques means that Talpa can find books and other media in ways that traditional search systems cannot.
