i knew about twitter for quite some time. bt then i didnt feel like letting the whole world know few stuf which i would say out to ma frnds! ;)
when i was thinking about twitter recently, and  how cool the concept is to jus say in lines wat th hell u r currently doing or upto!

something lil weird n insane struck me..

its like, when you say in english that you watched movie. the expression could be in words of your choice.. and when some1 of ur frnd says the same in english.. it could be same words or .9 prob with different words..  now if we need an engine to find the similar activities among your frnds, or among the folks in the online communitty, understanding the crap of english you all have written is quite a difficult task, which includes data mining and semantic analysis...

on the other hand, think of something qutie geeky like having a class for movie, and you say
new Movie("name of the movie", watched|wana watch | or fixed enums like this, "your comment on it!", may be rating.., and so on!)
Movie can have data members.. you jus fill things which you feel like..
when another frnd of yours does it, its easy to map em. .

you might get back to me, how th hell am i gona ensure folks enter correct movie name.. Movie name, could change to IMDB number, where entering the name does a query to IMDB and gets the imdb no.

All of us eat.. so its new Eat ( say wat u have ate) or its like new Eat ( new Hotel ( ) ) and so on!
you describe your current actions.. as semanticaly relevant objects..

i am sure we do stuf daily which repeat daily! Pumping lemma! i havnt bothered asking ma College Theory of Computation staff, who Lemaa is?! :P bt thn dont bother i wont ask u either :P ;)

if a single person repeats task daily.. then with th no. of folks in the community most tasks can be same or same tasks done differently..
so all i was thinking is, if we want to find folks doing similar tasks the same way or different way.. we could go fr having a initial set of tasks.. n let the users build the list of tasks..
a GUI can replace wat i mentioned about "new Movie()" stuff ;)  i wont mind th application giving a cool telnet theme where users write code to say wat they do presently ;) wat say! ;)

Do note, not that we need to limit users saying things tat are predefined activities, they can create new activities and still they can go n write their one liners as they do currently in twitter! ;)

Today one awsome luking gal was riding in an auto rickshaw.. seated in opp direction to th driver.. so tat riders behind th auto could see her.. ahhhh she was awsome!!!  as every story, even i had a Villain! th villain was th COPS! there came a police jeep in opp direction, WRONG LANE :X n th scooter felo riding next to me had to push closer to me n hit me, coz of police jeep.. :XXXXX some how managed to balance th byk.. Muah thanks 2 ma sw8 hr8 byk... bt i jus heard a imaginary voice.. some one calling me a MORON! its ma byk!! :P ;) 

Finally thought riding to infy (ma office) earliest was best choice rather following her! all left was the  lil pain in th foot.. whr th scooter fela bumped into me! :( :( :P

did i frget to say, tat the gal works in ma office.. one among th 10k ppl @ hyderabad branch :P bt still i never miss a single gud luking female in ma RADAR :D eh he .. bt all she scares me is by carrying  a damn lappy now n thn! Only PM's do carry around lappy's in ma company!!! Oh ma god! am js a DEV A.K.A CodeMonkey :P ;)

i was thinking of associating the contents of different web pages. Consider this example, i download movies ( i dont, if it comes to answer an attorney! am here blabbering to prove ma idea ;) ) from torrents. its possible to find the top 100 movies torrents. its possible to find the releases based on pirates. its possible to find the torrents based on the seeders. some times, i prefer watching the R5 release of movies. in this case, i prefer finding the top torrents of R5 release, which evidently becomes sort by seeders!
now identifying things based on seeds will not be perfect, i intend to find the movies IMDB scores. so i got to manually do a search ( google or live) to get the imdb page for th movie. then luk for the imdb score.
its like geting the data from mininova or piratebay then associating it with th data in imdb. ==> to refine the data to ur desires. in my case i had made the data flow btween two pages, in real world it could pass thro N pages, to get a final result. it can also invole N-1 intermediate results. which is also important.
i was surfing around to find, data extraction from tables to be popular.. bt then utlizing the extracted data to associate with another content.. n refining the search or data mining is wat i want. i feel its not out thr!..
thinking further, bt i jus wana jot it down in ma blog.. so tat i can keep track of ma thoughts ;)