Poolcasting

Claudio (from IIIA and sponsored by MyStrands) gave a very interesting talk about poolcasting (pdf of his slides). Poolcasting is a web radio in which individuals may join different channels. Those subscribed to a channel will listen to the same stream of songs. The problem is how to select the songs on that stream. Claudio did so by combining the preferences of a channelĀ“s listeners using Case-Based Reasoning.

The same approach may be used for mobile music. A bar may decide to play songs depending on the preferences of its customers (and preferences may be elicited from the playlists that customers store on their mp3 players or mobile phones).

A couple of questions that might be of interest to some of us: what if listeners do not share many songs in their playlists? Would it be possible to factor listenersĀ“ reputation (trust) in deciding which songs to play?

7 Responses to “Poolcasting”

  1. neal says:

    Well, not sharing your playlist might just mean that the chances of hearing something you like go down.. but factoring in reputation is an interesting idea. Would it be a reputation based on who provides songs that make the most number of people happy? It’s a difficult thing to measure..

    I suppose that for mobile music you may also need lots of people with similar tastes co-located in the one bar.. if not the only thing people may have in common is their love for the beatles! (http://www.last.fm/music/)

  2. Ops :-) By “not sharing” I meant: what if playlists do not have common songs (sparsity problem) :-)

  3. Well, since listenings are on a power law, the most popular bands (Beatles, Red Hot Chilli Peppers and Radiohead, according to last.fm) have the highest chance of being liked by everybody. In other words: when you have a heterogeneous audience, going mainstream is probably the best you can do (that’s why the market does it when the offer is limited, after all!).

    I wonder what would happen if this mechanism would be used: would people choose places where they like the music, creating a feedback effect that encourages specialization, or would all places converge to the same kind of mainstream music?

  4. That’s usually the case, but not always :-) It may well happen that guys in a channel may all like like jazz but their playlists have no common songs. In that case, one can’t go mainstream (cause the guys all like jazz), but one may still draw similarities among the songs in the playlists. How? Not sure :-) By putting songs in a spectral layout? :-) Or using something similar to this old paper
    “Horting hatches an egg: a new graph-theoretic approach to collaborative filtering”?

  5. neal says:

    if you already know that everyone likes jazz (because you make a genre-specific radio channel), then you can build the missing similarities using the feedback mechanism! so i guess the lack of pre-measurable similarities can be compensated for by getting real-time feedback. the only problem would then be if your listeners still all disagree as to what is good or bad (maybe that is why their jazz playlists have nothing in common)!

    .. thanks for the paper pointer, i haven’t read that one yet..

  6. They should do it also in clubs ;-)

    “Starting tomorrow at certain Starbucks stores, a person with an iPhone or iTunes software loaded onto a laptop can download the songs they hear over the speakers directly onto those devices. The price will be 99 cents a song, a small price, Starbucks says, to satisfy an immediate urge.”

    from: http://www.nytimes.com/2007/10/01/technology/01impulse.html?_r=2&hp&oref=slogin&oref=slogin

  7. [...] and poster session was very interesting as well: I saw a demo of Claudio’s poolcasting (see previous post) and so many posters that it is hard to write any kind of short summary of them. Conclusion: the [...]