Discussions on Book

Excellent idea. It should be pretty straightforward for me to translate the existing film award model to books.

It's working now. I've started you off with the 1970 Hugo for " Time Considered as a Helix of Semi-Precious Stones ."

ISBN is found in the "Edition" type -- books can have multiple editions, each with its own ISBN.

Under Edition an ISBN13 property would also be useful (possibly necessary in the future). Also, I don't know if there is a way to standardise ISBN entry, so that the check digit is either always calculated or always 'X'. It would make searching for a book / edition by ISBN more straight forward. To illustrate: The Catcher in the Rye has the ISBN 0316769533, but it could be entered 031676953X, as the last digit is a calculated check digit and doesn't actually carry any meaning. I thought I'd add to this - rather generic titled ISBN thread - rather than starting a new one.

Please ignore my previous comment - I just re-read the page on entering book information, ISBN13 is to be entered under the ISBN field. Although it would be nice to have an auto conversion from the 10 digit version to the 13 digit version.

Yeap, i have the same problem with Vargas Llosa.

This is fairly common problem, unfortunately. This was discussed on the Freebase data-modeling mailing list a little while ago, starting with this message.

More generally, this post on the Freebase blog discusses some of the ways internationalization is (and isn't) handled in Freebase.

In the short term, I'd recommend using the English title (when there is one) as the main title, and entering the original-language title as an alias. (If you're entering book editions, however, you should always enter the title as it appears on the book, no matter what language it's in.)

If you're feeling enterprising, and don't mind mucking about in MQL, you can try to insert the title in Spanish, following the examples in the blog post.

I'm afraid that currently, there is no good answer on how we should name books (and such) with non-english titles.  There's been discussion around this issue - see the data-model list for the thread. 

Unfortunately, that also does not answer your question at hand.  Aliases show up in autocomplete, so you can input either the original language title or english title as the alias and it will show up in autocomplete.

 Also, you can check out this excellent blog post about freevase and internationalization by crism.

freevase?  I mean freebase. ;)

The genre field lets you label a book as fiction or non-fiction. These aren't mutually exclusive, and you can add more than one genre to a book. The question about characters, though, doesn't have a perfect answer. Either we have the current situation, in which a lot of books have a necessarily blank field (i.e., non-fiction works), or we change the defaults and require a book to be typed as fiction before you can add characters. One semi-workaround for the blank characters is that you can hide empty fields, as per the link near the top of the page. It hides all fields, however, which may not always be what you want. Hope that gets a little closer to what you're looking for!

Whether a book has characters or not does not always correspond to whether the book is a fiction or non-fiction. Book Character as a type is not limited to fictions. For example, "A Beautiful Mind" is a non-fiction biography in which John Nash is definitely a (not to mention the main) Book Character. But because the character is, for all purposes of this discussion, "real", the type Fictional Character would not apply. Note the difference.

Oh, and it's OK to leave fields blank. :)

Right. I think the tricky thing here is that the Character property is intended for fictional characters only--though, as you note, people do use "character" to refer to people (and dogs and wookies and other sentient beings) in both fiction and non-fiction. The Subjects property is where we'd list real people, though there isn't anything stopping you from putting fictional characters in there, too. The good news is that the "What do I input?" info for these is pretty helpful.

I don't think the Character property is intended for fictional characters only; the expected type for that property is Book Character and not Fictional Character, which is a separate type in the Fictional Universes domain. The distinction I read from that is that non-fictional characters in a Book or Short Story would be typed Book Character but not Fictional Character, whereas fictional characters in a Book or Short Story would be co-typed both Book Character and Fictional Character. Isn't co-typing wonderful?

Speaking of which, I forgot to mention in my previous post that there is a "Work of Fiction" type that should be applied to fictions and can be used to distinguish them from non-fictions. The Publishing domain has an extensive array of types to help readers navigate the massive network of overlapping categories. For example, "Harry Potter and the Order of the Phoenix" would be co-typed "Book", "Work of Fiction", "Published Work", "Serial Installment", "Reviewed Work", and I hope, "Literature Award Winner" or at least "Literature Award Nominee." I'm a fan all right. :)

Hold on -- my mistake. Book Character has Fictional Character as an included type? Hmm, interesting choice. Why do we need the Book Character type then? ;) And how do I find all the non-fictional books that have a certain person who's not the subject? There's no way to query for that in Freebase?

I stand by my wish for co-typing "Harry Potter and the Order of the Phoenix" as a "Literature Award Winner", but it may be time to go home now that I no longer think faster than I type. ;)

All the various character types (book, film, tv, comic book, comic strip, play, opera, video game) have "fictional character" as an included type (or they were meant to; I'm too lazy to go verify that right now), because that's the most common case; you can always remove the type if it doesn't fit. If it transpires that non-fiction books are having significant numbers of characters entered, we should revisit that decision, though. There are two main reasons for having several character types -- one is so that every fictional character doesn't have a long list of properties ("books appears in"; "plays appears in", "comic books appears in", etc.) since comparatively few characters will actually have that many various appearances. The other is that a few of them (film, tv, theater, opera, and potentially video game) have additional properties that are specific to their domains and so require separate types. Having extra book, comic book, etc. character types allows us to be consistent. (HPATOOTP won both the Bram Stoker and German Phantastik Preis awards, you'll be happy to know -- feel free to add them!)

Thanks for the explanation, Jeff. HPATOOTP? That's one of the longest shorthands I've ever seen LOL! :)

I'm not sure about the sequel/prequel properties, since a lot of series aren't really ordered. Series that include both novels and stories often have an order for the novels, but the stories don't necessarily fit. But we absolutely do need a book series schema; I'll add it to my queue!

How should we address the overlap of data with fictional universe? Fictional universe has the property "works set here", which in many cases will be identical with a property like "books in this series".

I view works set here as a more 'universal' listing of items from all media and so is complementary to the more specific list of book series, film series, game series, tv series, comicbook series (superman is a classic example of all this).

So using the Freebase=iPod metaphor (in celebration of the iPod Touch launch) Fictional Universe could be the master playlist of all superman media and each media domain has a more specific subset playlist :)

Agreed; I think there's a definite need for this (as well as citations in academic articles), and it's something we plan to add in the future (hopefully the relatively near future, but we'll have to wait and see).

I agree: It could be interesting to add references of a book/article published in other books or articles and also online - something like what does Google Books. I'd also suggest: "related books" and "link to full-text" (different from the actual "link to web").

For e-journals I'd suggest as important information for authors and readers a metadata on open access. There is a sort of classification of publishers compliant to open access on Romeo Project website: http://www.sherpa.ac.uk/romeo.php.

What sort of information, or rather, what type of relationship would you expect "related books" to capture? "Link to full text" is a good idea; I had bee thinking the "website" property could do for that, but you're right that a specific property would be easier to use. I want to hold off on adding it until a new feature is implemented that will allow me to add the property to book, poem, story, etc. more easily and consistently.

As "related books" I intend related editions of the same book (by means of xISBN?), and may be also the books that share the same subject.
For "link to web" it could be interesting to link web sites related to the book - for instance the website of Wikinomics for the homonym book .
In addition you could keep the "full-text" for the online version.

I think we're already capturing this information, then -- editions are separate topics from books, and are linked from the book topic via the "editions" property; the "book subject" type already links the books (and, I think, essays) on the same topics. The related books are one click away, rather than being on the same page, but I think this should work well enough.

I've gone back and forth about 5 times in the last couple minutes about whether we should have a separate property for "related websites" or whether we should just use the default "web links" property. There's a case for either. I'll post a query on the data-modeling list and see what people think. Your additional thoughts on this would be very welcome, too.

The consensus on the data-modeling list is that the current "web links" property is probably sufficient for related web sites. The web links property accepts both a URL and a "title", which can be used for description -- "official website"; "fan website", etc.

Related books: mmmhhh... I'm not so sure that "one click away" is as having related books displayed in the same page... Have you had a look on books' pages in Google Books? This is what I mean.
What about xISBN? May be it could be useful for bulk uploading of bibliographical information and - once it is retrieved - it could be nice to see all in the same space.

Websites: I agree on "weblinks" + qualification in the title. That covers, however, only the sites of/about a book. I think it's important to preserve also the full-text link for all the digitized books that more and more public and private companies are putting online. Here a problem could be how to link books digitized by other providers, like Google Books. There is an interesting work going on by Librarything about this: http://www.librarything.com/blog/2007/09/google-book-search-on-librarything.php.

I definitely agree about online versions having their own links; I'll add it to book as soon as I get a second, but I'll wait on the other types (story, poem, etc.) for a new feature.

As far as related books go, I'm not sure. If it really is just books on the same subjects, we already are storing that information; to put it in another property as well means that users will have to query against both properties if they want a complete list. I could imagine someone writing an app that displayed Freebase books in a manner similar to Google books' display, though, which would be very cool. But I'm reluctant to put in duplicate data just because of the constraints of the current UI.

I added "online versions" to the book type. Thanks for that link to LibraryThing -- that's a pretty exciting project. Regarding xISBN, it would definitely be useful once we have a substantial amount of data; I didn't look into their licensing, so I don't know whether we could use it or not, but it's something we'll definitely look into.

Hi Jeff, it was so exciting to see the "online versions" link in the book record: WOW! :)
Now a problem might be the fact that some users have already put the link to online version of a book in the "Web link" metadata; I don't know if there is an authomated way to draw the value from "Web link" to fill the "Online version" considering that probably lots of "Web links" are there to intend "link to related websites" and not to the "online version" of the book itself... I'm trying to correct all the books I come across but I'm not sure I'll manage to cover all of them...

Related books: I'm not sure I have understood the problem, but if the point is: let's do an *interface* that gather all the info about the book, showing them like Google Books does, I think - optimistically :) - it's only a matter of time...

xISBN conditions:
"The xISBN Web service is free for non-commercial use when usage does not exceed 500 requests per day. [...]
The service is also available on a subscription basis for non-commercial and commercial use for usage that meets or exceeds 500 requests per day."
(from http://www.worldcat.org/affiliate/webservices/xisbn/app.jsp).

To me, it'd be interesting to use xISBN to *gain* "a substantial amount of data" (i.e. to upload bulk of books' records): you could query the OCLC database by ISBNs already on Freebase and pull down the related editions... Here comes another question I'd like to ask: I think it might be useful for users that want to create books in FB, have the possibility to search the main library opacs (online public access catalogs), as Librarything does - for example the opac of Library of Congress or the Worldcat one etc. The user could then capture the metadata of the book authomatedly, instead of fill in manually every metadata of the book ("copyright", "year", "subject", "Dewey Class" etc.).
I don't know, however, considering the structure of the books represented in FB, if in this case you can
1) extract only some of the metadata from, say, the Library of Congress' record, and put them in the FB *book's record* and
2) then put the other metadata in the FB *editions' record*.
(all in a way invisible for the user...)

In fact the record of a book in FB is about the *intellectual work* and doesn't contain all the info that, instead, are contained in the editions' records...

The LibraryThing model is a very interesting one, and is probably worth exploring further. It seems like there are some very good potential sources for book data out there, and I'm sorry to say that we just haven't had much time to sort them all out. We will get to it though! I just can't say for sure when. The difficulty of sorting the properties from one data-source into Freebase's system is pretty surmountable, though -- we've done it with music data from MusicBrainz, for one, and the music schema is almost as arcane as the book schema.

My guess re xISBN is that we probably qualify as commerical, but it's such a useful tool I'm sure we'll at least look into it when we get serious about loading more book data.

I agree: fortunately there are many sources to get records (and also full-texts) from out there. The last suggestion I would give you in this phase is to look also at Open Library, a project by Internet Archive to build the biggest free library on the net. This is the address: http://demo.openlibrary.org/.

I think the publishing domain is not that far off from the FRBR model, although obviously they don't coincide entirely (we have no equivalent for "item", for example). This is how I would interpret the "Divine Comedy" in FRBR and in Freebase:

FRBR:
Work: "Divina Commedia";
Expressions: the original Italian version; Mark Musa's English translation
Manifestations: "The Divine Comedy: Volume 1: Inferno" (publication of part of Musa's translation by Penguin); "The Divine Comedy" (Musa's translation as published in "The Portable Dante: Revised Edition"); "Lo purgatorio di Dante Alighieri Fiorentino" (a 1904 edition of the Purgatorio in the original Italian).
Item: The copy of Musa's "Inferno" that I read in my Medieval Lit class once upon a time

Freebase:
"The Divine Comedy" [for better or worse, we tend to use English display names] (types include: poem, translated work) [the abstract notion of the poem]
"The Divine Comedy" (types include: translation, published work) [Musa's English Translation]
"The Divine Comedy: Volume 1: Inferno" (types include: book edition) [the Penguin Classics edition of Musa's translation]
"The Portable Dante: Revised Edition" (types include: book edition, publication) [includes Musa's translation in the "contents" property]
"Lo purgatorio di Dante Alighieri Fiorentino" (types: book edition) [the 1904 Italian Purgatorio]

The reason that we can't use types to represent expressions/manifestations and have to resort to separate topics for each separate translation and publication, is this: Say we added a type like "published work" to the Divine Comedy topic, and this type was intended to contain information for every publication of the work (so it would have properties like "date of publication", "publisher", etc.). Each of those properties would have literally hundreds of instances: there would be hundreds of "dates published" and "publishers" but there would be no way to know which date(s) went with which publisher.

I think I have the point for what you say about impossibilty of using types to represent expressions/manifestations.
I'll continue to follow the developing of the project - it's so exciting for a librarian :)!