Problems? Is your data what you think it is? | |
PerlMonks |
creating crystal ball 2by grizzley (Chaplain) |
on May 19, 2008 at 19:56 UTC ( [id://687457]=perlmeditation: print w/replies, xml ) | Need Help?? |
Every time I have a while and go back to the problem, I am not able to think very much forward. Maybe there is nothing more to think about? So let's add few words and finish the topic. In the first part I wrote "All letters 's' at the beginning of word are marked bold. Two matches. Ok, we take first one always.". It definitely has to be changed. What options do we have here?
And Yet Another Idea to consider during implementation: mark with digits 10 best candidates in text with possibility to use via Ctrl+<digit>, Ctrl+<left/right arrow> adds/removes parts of text (words), Enter accepts. I was reading about LZW, Huffmann, Markov Chains, PPM, PAQ and other compression techniques suggested by you, guys. I don't exactly know how to use it to complement text. These algorithms operate on chars instead of words and this has made me think about following: suggesting one char is of no use. But what about suggesting 2 more chars every time user presses predicted char? Normal text at the beginning is what's already typed, bold is suggested(selection), and the rest light grayed to show what computer "has in mind". unemploymentunemployment unemployment unemployment unemployment unemployment Almost 50% of typing saved... Or better, 2 chars at the beginning, 3 in the next step, etc. unemploymentunemployment unemployment unemployment unemployment Well, so much for theory. I plan to start implementing it in a 2-3 weeks as I have now other important things to do, so please be patient :)
Back to
Meditations
|
|