As I read up on yesterday's hype of tagging it kind of struck me that nobody is using tagging as means to locate interesting content in p2p networks. There are quite some papers out there that try to elaborate tomorrow's p2p networks but the idea of tagging seems to be left out.
So here is my idea:
- Everyone canattach tags to his data.
- Since in most p2p networks data items are already identified by some hash. The tag is announced together with the hash and propagated throughout the network
- Clients will pick up the new tag, if they have the data item in their shares
- Searches will of course be extended to tags as well. So it will be possible to search for "news", "papers", "books" or some other tag or a combination thereof. This will allow meta-searches that are not tied to any "hard" data of the item such as filename, bitrate or size.
- Tags could even be created automagically by clients through existing databases, such as imdb or freedb.
- There is a high potential for improvements of course. Together with weighted filters, clients could sort out meaningless tag words leaving the interesting parts for the user to choose. (This is envisioned similar to the popular "'the' is a very common word [...]" Message known from search engines.)
That should be enough for now. This will be filed in the as-soon-as-I-get-time-to-implement-department.
If anyone knows wether this has Already Been Done (tm), please contact me through the usualy channels. If you are interested in Implementing this, feel free to do so.