One of the nicest bits of feedback we’ve had here at Last.fm HQ in the past few weeks was a message from two users who had just started a life together. They wrote to say that it all started through common musical interests on Last.fm, which led to them becoming friends online, and, over time, to discover that they were true soulmates. Yes, honestly.
As it happens we’ve been using the idea of a musical soulmate – someone who shares your most important musical tastes, your loves, likes and dislikes – to guide the development of our new neighbours service, which you can try right now, along with the other cool demos on playground.last.fm.
Our goal for this new service is learn how to introduce you to other Last.fm users who have lots in common with you – so you can be sure to find something you’ll like if you explore their home pages or listen to their personal radio. Making this vision a reality is challenging, but we think we’ve made a promising start. We’ve also tried hard to design the service so that we can quickly incorporate new ideas into the way we find neighbours – so please let us know how you think we’re doing, and what would make your neighbours even better.
So how is the new service different? In the first place we’re now able to use live charts to create a musical profile which we match against other users. This means that your neighbours can change as soon as you start to listen to new music. We think this is specially important if your tastes change over time, or if you’re new to Last.fm (we can start finding neighbours for you after the very first scrobble). In the second place we do more than just matching profiles. We make sure that all the neighbours we suggest have scrobbled something within the last couple of weeks, and if you’re heavily into several genres then we try to find at least a few neighbours for each of them. We take into account everything you’ve told us about your music listening, loves, bans, tags, etc. We also use some special stats to measure how aspects of your musical personality compare with other users. If we have time we may break out the coolest of these onto a separate “Compare yourself with…” page. In the meantime you might enjoy Anthony Liekens nice tool to compare tag clouds.
One feature we’ve already added to the playground demo is the chance to give us some extra help if you think we could do better, by suggesting some artists that your neighbours ought to be into, or not into. Even if you’re not that interested in neighbours, here in the research team we’d really like to hear how you get on, as we can use what we learn about neighbours to help improve all our recommendation services. And if you do find a musical soulmate through Last.fm, please let us know!