• What is this TwitterThing Mashup?

    Right now this is just me playing around with two different apis. Twitter and thingTitle from LibraryThing.

    • I came here from a Twitter Direct Message: What did you just do to me?

      I sent you a link to a recommendation for a book in LibraryThing based on keywords extracted from Twits in my friend’s list.

      • Why did you do that?

      It better be because you asked me to — by sending a reply with a “#gimmebook” in it. If not, accept my apologies right away, because I will be extremely embarrassed and upset about what is an obvious error on my part. Then, let me know right away!

      • It was a stupid recommendation — not relevant at all! Your algorithms are all messed up!

      Well, consider this a super pre-alpha gamma beta test. The code simply grabs words from my Twitter friends posts, chooses one random one and uses the thingTitle api on Library Thing to send you a link. It’s just for fun — a high entropy (but not random) way to discover a book.

      In future, I want to set up a mechanism for people to say “#gimmebook happy day” and I’ll send them a book related to “happy day” as well. (Incidentally, that would give you : Ruth Krauss’s Happy Day).

      • Why wouldn’t you use just my Twits. After all, those would probably make the book a bit more relevant to my day-to-day life.

      It’s possible and I’d like to, but the Twitter api currently restricts my authentication request to 70 an hour. Right now, running my script uses one request per launch. Running the script for an unknown number of people would make jump to one per user request. Not a big deal when you have your code running perfectly. But very restrictive when you are running code over and over again during the testing phase.

      • How does it work?

      When people send me a #gimmebook request, I launch a page that shows me the people requesting the book, what they said along with #gimmebook, and a LibraryThing URL and the word used to retrieve it. I check the ones I want to send a message to. And then I submit (automatically) DMs to those people that I approve to get the recommendation.

      • You do it manually? Why not just make it all happen automatically?

      Listen, I talk big, but I’m a neophyte coder. If I created a bot, it would probably try and take over the entire Internet. Bots could come later, but not until I’m sure the code is going to work the way I want it to.

      • Do you store information from my twits?

      No, I do not store them, nor do I have any plans to in future.  It turns out that I will have to temporarily store some status ids to make sure I do not send duplicate copies of your recommendations.   These ids are numeric, have no identifying info attached to them, and are only usable to people who have permission to follow you on Twitter.   Furthermore, they are only attached to your message which likely just says “#gimmebook.”

      I do see single words from your twits, and see them for a second as I launch the script, but if you are a friend, surely you do not mind me seeing one word from a large group of your twits mixed in with others, right? I also strip any word with an @ sign to prevent twitter logins from creating recommendations.

      But hey, if this really concerns you, I will not be offended if you de-friend me. It’s not as if I don’t already read your blog anyway.

      • How did you get this mash-up idea?

      Well, it all started with an idea from Amy Kearns. Setting up a Twitter reference service got me into the Twitter api. But deciding how something like that would work seemed hard from a people perspective making it hard to know where to start from a coding perspective. Then Julian and I brainstormed this idea. This idea was easier from a coding perspective, and therefore was a reasonable step toward what Amy wanted to do. That’s why I worked it through.

      • What’s the future?

      Well, LibraryThing’s api gives you ISBNs. If I have ISBNs I can make links to records in an OPAC as well (provided they offer an ISBN search). If all else fails, I can use WorldCat to do it. I may also take advantage of the TinyURL api to make some magic happen as well (many OPACs have looooong URLs).

      There’s also not a lot to show with this mashup, so I may see if I can get some of the bookcovers and make a fun “my twits in bookcovers” thingey to show it off.  If I do that, it will probably only be stuff from the public timeline, or my own personal timeline, though.