Book Dragon Data Sources

Data Sources

Book Dragon Archive uses several public bibliographic data sources to help identify books and fill in information about them. No single source has complete information about every book, every edition, or every identifier, and the sources do not always agree with one another, so Book Dragon checks different providers depending on what information is available and combines the results into suggestions for you to review. Before contacting an outside provider, the app checks your local library first, which means information you already have does not need to be looked up again. When an outside lookup is needed, Book Dragon sends the identifier or search information required for that request rather than uploading your entire library.

The three metadata providers currently used by Book Dragon Archive are Open Library, Crossref, and the Library of Congress. Each is useful for a different part of the identification process, and none is treated as a final authority simply because it returned a result. Book information is messier than it appears from the outside, particularly once different editions, publishers, identifiers, and cataloging systems begin describing the same work in slightly different ways.

Open Library

Open Library is Book Dragon's primary source for ISBN lookups and is also used for title and author searches, series catalog suggestions, and book cover images. The app accepts both ISBN-10 and ISBN-13 identifiers, and when an ISBN is available, Open Library is generally the first place Book Dragon looks for a matching record. Using it for several kinds of searches gives the app a broad starting point without requiring a commercial book database.

The tradeoff is that Open Library records can be incomplete, inconsistent, or ambiguous. A title and author search may return several possible books, different editions can be difficult to distinguish, and information that exists for one edition may be missing from another. Book Dragon therefore treats these results as candidates rather than assuming that the first record returned must describe the book in your hands. When there is uncertainty, the app presents the available suggestions for review before the information becomes part of your library.

Crossref

Crossref gives Book Dragon another way to find books, particularly in academic, technical, scholarly, and DOI-linked publishing where other sources may have weaker coverage. The app can search Crossref using a DOI, ISBN, or title and author information, which makes it useful when the identifier available on a book does not lead to a useful result elsewhere or when the publication is more closely connected to scholarly publishing systems.

Crossref contains much more than books, so receiving a Crossref result does not automatically make that result useful to Book Dragon. The app filters the records it receives and considers book-like material such as books, monographs, reference books, and book chapters while excluding journal articles from the book candidates presented to you. This does not make every remaining result correct, but it prevents a general publishing database from turning every identifier match into something Book Dragon assumes belongs in a personal book collection.

Library of Congress

Book Dragon uses the Library of Congress primarily when an LCCN, or Library of Congress Control Number, is available. The LCCN provides another route to a bibliographic record when an ISBN is missing, unsuccessful, or simply not the best identifier available for that book. Book Dragon also has a limited title and author search through the Library of Congress Books endpoint, which gives the app another fallback when the information available from a scan is less precise.

That fallback should not be confused with searching the complete Library of Congress catalog. A title or author lookup through the endpoint available to Book Dragon may fail even when the Library of Congress has information about the book elsewhere in its systems. In that situation, the failure means only that Book Dragon did not receive a usable result through the search method it currently uses.

Suggestions Rather Than Absolute Answers

Bibliographic information becomes complicated quickly once a book exists in multiple editions, formats, countries, and publishing histories. Two providers may disagree about a publication date, publisher, subtitle, author formatting, series relationship, or other metadata while still describing the same underlying work, and a provider may occasionally associate a valid identifier with an edition different from the physical copy being scanned. An external lookup returning a result therefore tells Book Dragon that it found a plausible record, not that the identification is guaranteed to be correct.

Book Dragon combines the information available from its providers into reviewable suggestions rather than allowing one source to become the unquestioned answer. When an ISBN is scanned, the app validates the identifier before searching for it, but a valid ISBN does not guarantee that one of the providers will have a matching record or that the available record will contain everything Book Dragon needs. If the lookup fails, the identifier can still be kept and you can continue entering or correcting the book manually. A missing database record should not prevent you from cataloging a book you physically own.

Scanning and OCR

The camera, barcode recognition, and printed-page OCR systems are separate from the bibliographic providers. The scanner may discover an ISBN from a barcode, or OCR may find an ISBN, DOI, LCCN, title, or author on a printed page, but Open Library, Crossref, and the Library of Congress are not performing the camera or text-recognition work. They receive the useful identifier or search information after Book Dragon has determined what it should look for.

Printed-page OCR runs on the device, and the page images and raw OCR text used during that process are not stored as part of the metadata lookup flow. The purpose of the scan is to extract enough useful information to identify the book and perform the appropriate lookup, rather than sending the scanned page itself to the bibliographic providers. Book Dragon also checks the local library before beginning those provider searches, so the presence of external metadata sources should not be understood to mean that your book collection is being uploaded to them.

Working with Imperfect Data

The current providers give Book Dragon access to a large amount of useful book information without making the application dependent on an expensive commercial metadata service. That choice is part of the same architecture behind our goal of keeping Book Dragon Archive's primary functionality free forever, but it comes with a real tradeoff: public bibliographic data can be incomplete, inconsistent, oddly organized, or simply wrong. Adding more sources can improve coverage, but it does not automatically clean the information those sources provide.

Archive Nine is developing its own book information API to help reconcile some of those problems, particularly the relationships between editions, series, universes, and genres. That system will develop gradually because useful book data has to be connected and cleaned rather than merely accumulated, and we do not intend to make it the single service Book Dragon depends on to function. The public providers will remain part of the lookup system so the application can continue doing its basic job even when our own service is unavailable, while our API can increasingly help make the information returned by those sources more coherent.