Ideas:
1. People need to be able to add their board, and these should do to a manual approval list.
2. Need to set link depth, as some boards are open flat threaded (popbitch, rathergood) and others are clicky (popex)
3. Possibly you should think about setting WHEN the poll is done. Some boards may need polling every couple of hours, whilst others move more slowly.
Boards to poll to start with
www.b3ta.com/board/
www.popbitch.com/board17.html
www.urban75.net/vbulletin/forumdisplay.php?forumid=6
(,
Fri 8 Feb 2002, 6:16,
archived)
1. People need to be able to add their board, and these should do to a manual approval list.
2. Need to set link depth, as some boards are open flat threaded (popbitch, rathergood) and others are clicky (popex)
3. Possibly you should think about setting WHEN the poll is done. Some boards may need polling every couple of hours, whilst others move more slowly.
Boards to poll to start with
www.b3ta.com/board/
www.popbitch.com/board17.html
www.urban75.net/vbulletin/forumdisplay.php?forumid=6
most message board content isn't so much links to things as general discussion of a topic.
That said, discussions often start, or are soon followed up with, a link to some other material. In this case, the link is probably to old stuff, but the discussion is more new/interesting.
You will need to record the parent of any link you find, ie the route you took from the main board to the page with the link, so you can tell when that has dropped off the front page. This will give you a measure of how often the board updates.
Link depth can probably be determined by seeing if all the links (well *most* of em)on the page point to deeper down the same branch of the site. Or something.
This is fun. Wish I had the time to experiment... but I've got to finish my automated Win2K install script.
(,
Fri 8 Feb 2002, 6:38,
archived)
That said, discussions often start, or are soon followed up with, a link to some other material. In this case, the link is probably to old stuff, but the discussion is more new/interesting.
You will need to record the parent of any link you find, ie the route you took from the main board to the page with the link, so you can tell when that has dropped off the front page. This will give you a measure of how often the board updates.
Link depth can probably be determined by seeing if all the links (well *most* of em)on the page point to deeper down the same branch of the site. Or something.
This is fun. Wish I had the time to experiment... but I've got to finish my automated Win2K install script.
... b3ta/popbitch gets lots of links. And b3ta gets most of it's clickthrus to the viral blobs via other message board traffic. I'm interested to monitor link popularity via message boards, as the barrier to entry is lower than to blogs, I think the links might be more representative of what people actually are looking at.
You could do some really funky stuff with google groups... with these ideas.
(,
Fri 8 Feb 2002, 6:46,
archived)
You could do some really funky stuff with google groups... with these ideas.
more on lines of the loons in the self help groups (amonst many others)
(,
Fri 8 Feb 2002, 7:48,
archived)
Now flat boards are relatively easy. just take the current page and update it in the index.
Proper threaded ones are going to be a little harder - not so much the indexing of the content, but
deciding what content to index.
To get the most "current" board topics we need to index the "most current" threads, which theoretically
are those which have the most recent posts.
The question is do we want to index the current content of a thread
1. - (i.e. the newest posts to the most recently updated threads)
or just the original content of a thread
2. - (i.e. the original content of the most recently updated threads)
I guess point 2 will give a better idea of the context of a thread (maybe), but point 1 is a lot
fresher....but with a risk of indexing posts that just say "yeah, I agree".
Also, is the primary aim of the service:-
- To provide a searchable index of several bb's
or
- To provide a list of most current discussion subjects across all bb's
Planning to do both, but back-end data will be structured differently according to what we want it
to do.....
BTW, If this post is getting in the way of the usual ramblings we can always move it to an email
discussion...although that's less fun (and generates less ideas)
[email protected]
(,
Fri 8 Feb 2002, 7:08,
archived)
Proper threaded ones are going to be a little harder - not so much the indexing of the content, but
deciding what content to index.
To get the most "current" board topics we need to index the "most current" threads, which theoretically
are those which have the most recent posts.
The question is do we want to index the current content of a thread
1. - (i.e. the newest posts to the most recently updated threads)
or just the original content of a thread
2. - (i.e. the original content of the most recently updated threads)
I guess point 2 will give a better idea of the context of a thread (maybe), but point 1 is a lot
fresher....but with a risk of indexing posts that just say "yeah, I agree".
Also, is the primary aim of the service:-
- To provide a searchable index of several bb's
or
- To provide a list of most current discussion subjects across all bb's
Planning to do both, but back-end data will be structured differently according to what we want it
to do.....
BTW, If this post is getting in the way of the usual ramblings we can always move it to an email
discussion...although that's less fun (and generates less ideas)
[email protected]
I was interested in polling for links. I think there is value there.
(,
Fri 8 Feb 2002, 7:28,
archived)
is a fairly quick hit I think - and doesn't require too much storage, so I'll do that first.
The question is do we want link and link text or link & sample of target url....
[The more complicated, storage instensive, thread archivy thing may offer a
more detailed take on current discussions tho (so I might build that later...)]
(,
Fri 8 Feb 2002, 8:21,
archived)
The question is do we want link and link text or link & sample of target url....
[The more complicated, storage instensive, thread archivy thing may offer a
more detailed take on current discussions tho (so I might build that later...)]