|
Does there exist an ontology for linking people with tv/radio/theatre shows they have watched/listened to? It would be useful for recommendation engines to know this so that they don't recommend something that you've already seen/heard. I've been thinking about writing an audience ontology myself, but was wondering if anything similar existed already. |
|
Seems to be that it might be useful to define not a full ontology but a single property "observer" with domain http://purl.org/NET/c4dm/event.owl#Event and range foaf:Agent. Both the programmes ontology http://purl.org/ontology/po/ and music ontology http://musicontology.com/ subclass cd4m:Event for their po:Broadcast and mo:Performance classes, so such a property would be compatible with two very relevant ontologies. I was thinking of doing something similar to this. But for the example of recommendation engines, people don't really care about which po:Broadcast or mo:Performance they saw/listened to, they're more concerned about which po:Episode or mo:MusicalWork it was. However, from the broadcaster/venue's point of view the po:Broadcast/mo:Performance is important (for viewing figures etc.) A po:Broadcast can be a blank node - e.g. <#some_episode> po:version [ po:broadcast [ ex:observer <#me> ] ] . Of course, it might be useful to define a shortcut property for uses like that. |
|
There's some exports/properties exported from online music sites. Some properties I found:
Other than that, I don't know of one. (IMO, such an ontology would be very useful.) |
|
I guess the Play Back Ontology should fit your use case. There you can model Play Back Counter, which are in relation to ScrobbleEvent instance. Thereby, scrobble event represent a general multiple purpose activity record event. That means, if the play back counter concept is to specific for some of yoru use cases, you can also switch to the more abstract modelling represented by the Counter Ontology. Another option is the Cognitive Characteristics Ontology, which is an enhancement of an unioned interests ontology, namely the Weighted Interests Vocabulary, on top of the mentioned VoCamp proposals from Thomas Francart. Finally, if you like to model high-level recommendations as Semantic Graphs, I can recommend ( ;) ) you the Recommendation Ontology. PS: I also like Toby's "observer" proposal, but I guess one would end up in a property reification ;) |
|
You might want to have a look at the "User Weighted Interests Ontology" : http://vocamp.org/wiki/HypiosVoCampParisMay2010#User_.28weighted.29_Interests_Ontology You'll find here pointers to other "user-interests related" vocabularies. |
Thanks for all the responses, I've written up some of my ideas in a blog post http://blog.lukeblaney.co.uk/post/1294635332 - any feedback would be much appreciated.