Hi everyone! It's been a while. Same excuses - lazy, busy, too much Dota, you know how it is. I hope.
I've also made some progress on my corpus study tool. I've got the framework for discovering probabilities of particular sequences, as shown at left. To explain this, what we have are sequences of words (with an intermediary separator, don't worry about that) in pairs, followed by a final word and the number of times it occurs after the pair. You might also notice I'm clearly using The Call of Cthulhu as my test corpus. The next step is to aggregate by initial words and then make a list of all the possible finals with their counts, and then finding a final probability of some sequence of three words ending with a word can be determined pretty trivially.
So anyway, I've been doing some things. Among other stuff such as guitar and Halo 4, I've been working on my code projects and made a bit of progress on some stuff! Let's get to it.
As you might recall, my chess clone, last time we visited, was completely stupid. The pieces knew where they could move, but upon trying to do so they made a pawn where the move would go. There were a number of pawn-based revolts on the basis of their value depreciating from the influx of pawns in the market. But it's fixed now, as you can see. Yes, there are also moves that shouldn't be there off the board, but I'm still working on that (if you're curious, it has to do with capture moves being a little wonky with my setup).
So there's that. It's one step closer to a functional clone that I can actually do stuff with, and that's what matters when I can't commit to this stuff full-time.
I've also made some progress on my corpus study tool. I've got the framework for discovering probabilities of particular sequences, as shown at left. To explain this, what we have are sequences of words (with an intermediary separator, don't worry about that) in pairs, followed by a final word and the number of times it occurs after the pair. You might also notice I'm clearly using The Call of Cthulhu as my test corpus. The next step is to aggregate by initial words and then make a list of all the possible finals with their counts, and then finding a final probability of some sequence of three words ending with a word can be determined pretty trivially.
Not sure what I'll do with it once I finish that step (which might happen tonight), but it'll be more or less complete for English/Latin symbol languages sans doing a bunch of stress testing on more Lovecraft and making it not an aesthetic mess. Then I'll have to work on a method for finding word boundaries in languages such as Japanese or Chinese, which will be a huge mess. Looking forward to it!
The last project I've been working on is the guitar project, to which I haven't added much. I just made it possible to do lookups on multiple notes and find scales containing all of them (shown at right). I've also been trying to figure out how best to implement chord lookups, which in and of itself wouldn't be too hard, but I want to be able to find chord variations (like an Asus4 or Gm7 or however the spirit moves you) and show them (and thus, all possible positions) on the virtual fretboard. I'll also at some point need to start making this guy graphical.
So, that's what I've been doing, as well as Minecraft and Dota, and I'm probably going to use my 2 free months of XBL fairly soon to kill scrubs in Halo, if anyone ever wants to join me on that just lemme know. Until next time!
No comments:
Post a Comment