When you filter out useless info, generally you look for words you do not want. This is easy, but sometimes we miss out on things we actually want. And these could actually be important. But it is relatively easy to do.

However, if we try to collect only the things we are interested in, the quality of content would be generally improved, and true negatives would be reduced. The difficulty is with generating the list of "interesting" keywords.

One big problem I have been facing is the lack of a good filtering method. For example, If I am not interested in stuff about the Palm Pre, I should do the clever thing and block it in the Y!Pipes already. However, it could be involved in a parody of something else that I am interested in, or just simply related to some other field I am interested in, I would not receive that info.

So now, how to generate that list of keywords that I actually want, without sitting down and writing down every single thing? I can think of something related to Bayesian theory, like I choose stuff I'm interested in and it starts to learn, but I have no idea how at the current moment.

Or I can stick with status quo and spend 15 minutes per 45 minutes weeding out the not so interesting stuff from feeds. Not very fun.

0 comments: