I think we can calculate ratings based upon the content of all the RSS feeds. We should be able to use the link or guid tag to determin the uniqness and source of a peice of content. We can keep running totals of things and use some sort of equation or something to determine the rank which will help admins find the content that is more likely to be good so it can bubble up.
-Neil ---- Original message ---- >Date: Fri, 18 Jul 2003 01:35:07 -0700 (PDT) >From: Ka-Ping Yee <[EMAIL PROTECTED]> >Subject: [hackers] Syndication, bubbling, and rating >To: <[EMAIL PROTECTED]> > > >| <Zacker> my idea is we figure out a weighting system >| <Zacker> that can assign a number to every piece of content >| <Zacker> like page rank >| <Zacker> 1 to 10 or something of the like >| <Zacker> and that number is calculated by >| <Zacker> how many people syndicate it / view it or something like that >| <Zacker> basically its popularity rating >| <Zacker> and it meta dean keeps track of it.. and this is used in the >| "bubble up syndication" stuff >| <Zacker> so admins can flag feeds to watch for high ranked content >| <Zacker> to auto syndicate >| <Zacker> or flag for them to look at... or the like.... > >I don't think this is practical (at least in this form). The >description suggests that ratings for *every* piece of content >will be housed all in one central database -- and that this >central database must be updated whenever *anyone* reads or >syndicates any piece of content. We can't have all the hundreds >(thousands? How many are we expecting?) of sites out there >constantly hitting our database with updates. > >I'm not saying ratings are totally unrealistic, just that a lot >more architecture would need to be designed to support it. > >Key Question: What kind of scale are we expecting? It would be >good to have rough estimates for: > > - number of sites > - number of users per site > - number of articles per site per day > > >Sorry to keep nagging but i'm going to raise this question again >because IMHO we have to make this critical design decision in >order to proceed: > > * What determines the set of article pointers that each > site will cache? > > >And a related question: > > * How does content "bubble up"? Will parent nodes have to > poll all of their child nodes to look for content to promote? > > >-- ?!ng >
