FoafInGnutella
From FOAF
The WorldWideWeb is not the only system on the internet which can benefit from RDF.
p2p networks could greatly benefit from following the foaf:knows properties when connecting to nodes.
foaf:knows is used to designate people you know. The relationships vocabulary is an added layer atop that framework, but the most common purpose (as I see it) is "I am a friend/associate of Person X".
People tend to flock together like birds of a feather. I like certain types of music, so will some of my friends. Interests bind us together.
P2P tries to be a massive network - but it fails where a simple one to one server client connection would succeed often. If I know 3 people with rare files, why can't I direct my client to swarm from them instead of wasting its time chasing ghosts at random with low distribution rate files.
Much hubbub goes on amongst the p2p community about improving the network - but most are ignorant of the Semantic Web. p2p often fails because my peers are too inspecificly judged to be my peers. Anyone will do when I connect, because the message gets their eventually doesn't it?
Using the power of mapped out social networks to navigate through p2p could mean significant increases in performance for p2p. If my friend wants to send me a video, he can send it direct or give me a hash to search for. Why can't he skip out the search element and just start sending me chunks ASAP? If he and all of his friends, rather than a random assortment of nodes around the world, are directly connected to me, I can get what I want sooner. Geographical convience is another benefit - local servers because 50% of my friends are going to live within the same country.
The widespread distribution of p2p applications means we could get FOAF descriptions of people happening much sooner and on a far more widespread scale. LeighDodds has authored a beta desktop FOAF-a-matic, this kind of thing would be a much wanted (people always want to add chat to p2p, but why not just add foaf:msnChatID?) feature to p2p clients. -- DanielOconnor
I'm having a hard time understanding what you are saying.
Are you saying: "P2P networks can search more efficiently if they start out by checking my friends, and my friends' friends, or people with similar interests to me, than if they just start searching out at random."..?
I'm not sure what you mean in the "if my friend wants to send me a video" case story.
As for geographical convenience- I don't know if it's that great; I think most people have most of their Internet contacts outside of the country! Also, geographic nearness doesn't usually translate into speed anyways. If your neighbor has a modem connection, but the person across the Atlantic has a DSL connection, you'll probably get the data faster from across the Atlantic.
When people are swapping songs, I think they want to be not personally identified, since most song swapping is illegal. But for the portion of people that want to swap legal songs, and other legal files, I think it's a great idea. :)
-- LionKimbro DateTime(2004-06-12T16:40:21Z)
You are right about most not wanting to be identified. This is why its a good thing for p2p networks - it would discourage illegal file sharing and help leverage the power of distributed file sharing for more legitimate purposes. Geographical proximity. You have a point there, but it saves unneeded relay clutter on the internet if you try your local guys first. What am I saying? "P2P networks can search more efficiently if they start out by checking my friends, and my friends' friends, or people with similar interests to me, than if they just start searching out at random." Thats it in one. I know that p2p networks index a file then find the server responsible for that part of the network, but wouldn't it be faster if there were smaller servers with which you enjoyed a more static connection with? If my friend wants to send me a video, he forces me to search from my nodes on in until I find the video - whereas he already has the file, and a collection of sources recorded who he got it from, as well as friends who also have it - so you don't have to search, its already been done there somewhat.
--DanielOconnor DateTime(2004-06-13T02:53:00Z)
