mahout-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ted Dunning <>
Subject Re: Evolution of ratings over time
Date Mon, 01 Oct 2012 01:03:31 GMT
Other experiments have shown that 60-80% of perception of music "likes" is
due to social factors.

Factoring this out may or may not be a good thing.  My feeling is that if
you are trying to make people happy with what you recommend then you need
to go with whatever makes them happy whether it is innate or not.

My favorite example of this is Bob Dylan.  If I (or anyone) likes his
music, it is highly likely that this is socially mediate more than
acoustically.  Thus if you recommend music that "sounds like" Bob Dylan,
you are likely to make people very unhappy.  On the other hand, if you
understand the social currents about which social phase of Bob Dylan the
person likes, then you are likely to make very good recommendations.

On Sun, Sep 30, 2012 at 11:13 PM, Lance Norskog <> wrote:

> Alledgedly, ratings have a "peer pressure" effect, and so earlier ratings
> are more valid. Your mileage may vary.

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message