Choosing for a trained highlight to affect a whole folder / Everything


I was wondering to which extend it’d be possible to train a highlighted word/phrase to affect a wider spectrum then just one specific site. If it’d be for example possible, after I have selected a specific word from a stories title to be thumbed up, to then choose whether I want that word to apply to “Everything” or to all the sites of a specific folder.

Case in point: I have a folder for science news from a couple of big sites. They throw all type of news at you, but I mostly don’t want to miss a certain type, say news about archaeology or linguistics. Same scenario in my general news folder. I have some major news websites added, which I will generally skim through if and when I have time. However I don’t want to miss their news on, say, my country of origin. Atm I have to train every one of the sites in that news folder individually to accept specific words, and that only if I come across the word in a title. The process is daunting, especially when first adding a site, and especially for those “general interest” massive websites. I am wary of adding all the news sites I want until I have sufficiently trained the ones I have, cause I’ll bury myself in yellow if I do. If I could however apply a specific keyword I just thumbed up to the whole folder, that’d save me a lot of trouble.

Training will become easier if you later implement the already suggested idea of adding a training option straight into the site addition process, but still, there are many occasions when being able to widen the scope of a specific thumb up or down would really help.

Is this even feasible? I seem to remember you giving reasoning against adding generalised keywords, but what I am suggesting seems (to me) closer to the way Newsblur already functions. Is it?


I’d say you’re right, this is an interesting idea. The problem is not applying the intelligence filters on multiple sites at once. The issue is keeping track of the filter, which would require a much more advanced dialog that exists now. And intelligence training is already complicated enough.

Actually, I’m in search of a way to simplify it further. So adding this kind of complexity wouldn’t work unless I could figure out a neat and simple way of applying intelligence classifiers to a folder.

The backend changes are a bit more complex, since it would have to watch the contents of a folder for changes, something that the current backend doesn’t support very well. But I could always hack it to work and it would be fine. I just need to figure out that front-end component and it hasn’t hit me yet.


+1 to this I would really like this feature :slight_smile:

Keep up the good work!


Same. +1 for the folder training


Or would it be possible to import the saved words from a feed to another one?