Regarding the “lafnlab/redirect-to” tag. We’ve started to see a convention where people use a “related” tag. This has taken several forms:

ntoll/related
ntoll/related-objects
ntoll/related-books

So the rule of thumb is to use related-foo where the “foo” qualifies what the relationship might be. If you just want to say something is related to something else then use just “related”.

However, I really like the semantic content of “lafnlab/redirect-to” since it shows that you think I’m looking at the wrong object. In itself, this is valuable information. I think I’ll start using this as a convention too (thus demonstrating another cool way in which Fluidinfo is “social”).

This is great stuff. Keep it up!

PS You should join us in the #fluidinfo chan on Freenode IRC.

Lately, I’ve been thinking more about possible projects that might come of this competition entry. When working on the entry, the idea was to make a site that pulled the basic (fair use) info about the books from Fluidinfo and list them alongside their other versions (i.e. foreign editions, audiobooks, etc). This wouldn’t just be for O’Reilly books, but for all books. The oreilly.com info would be a starting point, but the idea is to potentially include all books (thinking big). Then came the idea about a search engine for books. I’m still not sold on that idea, but the ideas are very fluid (no pun intended) right now.

There are places to find info on books, but they’re walled gardens, and Fluidinfo is aiming to make some sets of data more accessible to everyone – sharing the database.

The starting point of the project would be for me to code a webpage that queries Fluidinfo for books. Earlier today I used Fluidinfo Explorer to do this. I entered fluiddb/about matches "book", but this was imperfect. While there were a lot of books, too many of the objects that were returned were urls to pages at GoodReads, BoingBoing, or other sites. The project would have to have some way of filtering the results. Ideally, it would only return the objects that follow the pattern book:some book title (some author). After the object is pulled, the code should also pull look for tags such as: isbn, publisher, copyright, etc. Maybe if this data was missing, there would be some way for me to flag it – to either query Fluidinfo at a later date, or to tell me (or someone) to search for more info (kind of like the stub flags in Wikipedia).

However, Fluidinfo doesn’t seem to have that many books at the moment, so I would try to code another page that allows me (or maybe anyone???) to enter book objects. This is separate from the tags for isbn, url, title, publisher, etc. This page would strictly be for entering book objects. This is good if there are only a few books, but if there are dozens or hundreds to enter (say in a csv or tsv file), then it would be better to make some sort of script to enter this info. I might tinker with that in the coming weeks.

One possible problem with this is if there is an object that already has that data (an object with that name already exists). Presumably, Fluidinfo will return an error in that case, and the script would have to take the error into account.

Since I’m big on multilingualism, any project would probably sniff the USER_AGENT string and localize to the user’s language.

We made the API without input (or data) from O’Reilly. Only when we were done did we really tell them about it :-) I’d been saying for a long time that we were going to do it, but the thing which got the attention was to finally say “ok, it’s done”.

Thanks for entering the competition. I think you’re onto a good idea as the non-English speaking/reading segment of the world seems to be quite large :-) Being able to browse O’Reilly titles and see links to foreign editions would be great.