Pandora for Poetry

1 November 2006

Wouldn’t it be great if a service like Pandora existed for poetry? I haven’t peeked under the lid, but from the description Pandora seems to analyze relationships between musical albums using specific elemental criteria to describe and correlate the sound and feel of one band with another (e.g. fast drums in this one, fast drums in that one–one point toward possible correlation).

Could we not similarly bring the tools of critical analysis to bear on poetry in attempt to accurately describe a web of interrelationships between bodies of work? And then make that web navigable, both for critics to analyze for relevance to their research and for fans of poetry to discover new favorites? What would those defining elements be? And how would such a tool affect our understanding of poetry? (And our understanding of the possibilities and limitations of classifying art?)

Liked this? Receive new ones in your inbox.

(You can unsubscribe any time.)

Posted in Insights, Poetry, Technology | Comments (14)

http://www.pathogendetect.com Peter Gallant

Robert – wouldn’t “Pandora for poetry” run into the syntax vs. semantics issue? There are probably a lot more metrics that can be extracted from a song (digital signal processing, lyric, gendre and artist information available in nice, clean tags) than performing semantic analysis on a poem. Poetry is an artform that is subject to interpretation, has an individual meaning based on each person’s unique life experience and background, and often uses language in ways that would blow the electronic mind of any semantic analyzer.
Regards, Peter
Robert

Hey Peter,

Maybe. I’m inclined to think that literary criticism may have evolved enough tangible, discrete and even quantifiable “genomes” of poetics to allow real critics to sit down with a poem and a matrix and catalog it appropriately. By appropriately I mean enough so that it serves the end purpose of being able to recommend new poems that real people will really like based on those people’s current favorites. Then, of course, the question is: would that give us anything academically useful as a result? I think so. But I don’t think we’d know until we got there to some extent.

Certainly, I wouldn’t expect a natural language parser to be able to analyze a poem. That’s like trying to do OCR on a psychedelic CAPTCHA drawn by monkeys in a centrifuge — two few of the rules of recognition are obeyed to make it work. Funny to even think of a machine trying to parse, for example, some of the modern language poetry floating around. It would be confused as I am about most of that stuff. Then again, I doubt Pandora uses DSP to catalog its music. More likely, real people fill out a matrix. Defining that matrix, however, would be where the genius lies. Get it right, and it could further our understanding of art. Maybe.

Cheers, Robert
http://erif.org/ kaolin fire

I’m fairly certain pandora is doing all of their genomic tagging manually, if that helps anything in the thought processes.

It’d be interesting to see what matrix folks would come up with to start with. I’m definitely not poem-knowledgeable enough to really even begin to help with that sort of definition.

Folks on poemranker came up with their own matrix, as… well, more of a useful critique thing (and for lack of anything better to do)–but I don’t think it could be used to traverse the net of poetry.

Still, a nifty idea, and I salute it and would definitely spread word if you were going to attempt such a thing.
Robert

The implications are interesting. But for now, it has only made it as far as my someday/maybe list.
http://erif.org/ kaolin fire

I definitely know how that goes
jenni

Hi Robert, you won the wine glass contest. Send me your addy and I’ll mail the book out to you. Congrats!
Robert

Wow! That is so awesome. I went back and forth and back and forth. How close was I? I’ll send you my addy by email, since, y’know … I did my undergrad studies at a school that still had notices posted about the Unibomber.
http://www.pagehalffull.com/humanyms/ Pearl

That would be cool. It could input all the themes and flags like allmovie.com does. If tagged manually it could be PhD project.

What was the last count 10,000 publishing poets in America alone. How many poems? Which ones? How far back in history? How wide in geography? More than exclusively English? That would be fun to plan out at least.
Robert

I agree on the fun-to-plan factor. What do you think the “genomes” of poetry are or could be?
http://bloggingpoet.squarespace.com Billy The Blogging Poet

Well I can’t claim it to be anything like Pandora but http://www.poets101.com is the world’s first, and so far only full=featured poetry aggragator allowing poets multiple ways to gain exposure for free. We suffered a lot of technical issues at first but everything is full speed ahead now.
http://erif.org/ kaolin fire

Hey, nhice job, nice idea with that. Want to hook poemranker.com into it?
Robert

Thanks for the resource. You’re already getting offers to collaborate!
http://www.pagehalffull.com/humanyms/ Pearl

Billy’s been here. Poemranker looks interesting too.

Particular forms, or schools, or geographies, all coastal everywhere, or poltical or by shape of landscape, regions, descent from certain formative teachers who spam dozens or hundreds of emulators.

btw I apparently have thought along these lines before.
http://www.pagehalffull.com/humanyms/?p=266
http://www.reflectionphotography.biz ashley brook smith

I would love to see pandora poetry, because i think that there are a lot of poets out there who either aren’t sure what to do with their material or have trouble finding their nitch in life. Pandora poetry could help artists who are going through similar issues connect. Poetry is a lot like music, it even has rythm. If I were to do it I would either classify it by subjects or individual poets. Maybe the level of depth. It could possibly open up the world to a lot more people.