3 ms·

I really don't get what it means to "sort" the text. Can you describe that further? Aside from that question, what you describe sounds a bit like the online se

I really don't get what it means to "sort" the text. Can you describe that further? Aside from that question, what you describe sounds a bit like the online service SpinSpotter (which I think is a good idea, but severely underused).

DanielBMarkham 17y ago
The old idea was to "rank" something. You voted up or down to indicate -- beats me -- you liked it, you wanted other people to see it, you like the person, you're in a good mood. Whatever. Instead of a single dimension, my idea is that internet content consists of two things: keywords and emotional impact on the reader. So you tag the content, giving it semantic value, and then click on various indicators showing how it affected you emotionally. So now I can provide you with associated material that is relevant and can predicatively adapt to your moods, i.e. responses. It's cool stuff, but basically it boils down to a souped-up version of PageRank-like concepts. I have a screen shot of a UI prototype but have to catch a plane right now. Glad to put it up if anybody would like to see it.
- CWuestefeld 17y ago
  Ahh, I get it. The user isn't actually sorting, but providing input values for a segment of text, that a sorting algorithm might use. I was having visions of trying to parse sentence breaks, and then having users shove the good sentences in the article to the top, and bad ones down. None of which made any sense to me...
  - DanielBMarkham 17y ago
    Yep. So if the last three times you read something about Michael Jackson it ticked you off, the system doesn't rank segments of text tagged with Michael Jackson very highly the next time you search for related questions or statements. To take it a step further, if you and I track closely when the keyword phrase is "Sundance Film Festival", and I just read an article I found hilarious, it's probably something you would enjoy as well. There is a bit of arm-waving going on here, I admit. But it's just a blue-sky concept right now, so a little arm-waving can be forgiven (hopefullly).

DanielBMarkham 17y ago

The old idea was to "rank" something. You voted up or down to indicate -- beats me -- you liked it, you wanted other people to see it, you like the person, you're in a good mood. Whatever. Instead of a single dimension, my idea is that internet content consists of two things: keywords and emotional impact on the reader. So you tag the content, giving it semantic value, and then click on various indicators showing how it affected you emotionally. So now I can provide you with associated material that is relevant and can predicatively adapt to your moods, i.e. responses. It's cool stuff, but basically it boils down to a souped-up version of PageRank-like concepts. I have a screen shot of a UI prototype but have to catch a plane right now. Glad to put it up if anybody would like to see it.

CWuestefeld 17y ago
Ahh, I get it. The user isn't actually sorting, but providing input values for a segment of text, that a sorting algorithm might use. I was having visions of trying to parse sentence breaks, and then having users shove the good sentences in the article to the top, and bad ones down. None of which made any sense to me...
- DanielBMarkham 17y ago
  Yep. So if the last three times you read something about Michael Jackson it ticked you off, the system doesn't rank segments of text tagged with Michael Jackson very highly the next time you search for related questions or statements. To take it a step further, if you and I track closely when the keyword phrase is "Sundance Film Festival", and I just read an article I found hilarious, it's probably something you would enjoy as well. There is a bit of arm-waving going on here, I admit. But it's just a blue-sky concept right now, so a little arm-waving can be forgiven (hopefullly).