(31 Mar 2015 at 22:45)
As usual for this time of year, I've been hard at work on SIGBOVIK projects. The conference is tomorrow and I'll be emceeing (something I seem to always get asked to do, though I'm also happy to pass this torch soon, since what I really enjoy is making and presenting weird "research"). Speaking of weird research, I have two papers this year; the first one I spent weekends on for a month or two. It's an AI/machine learning joke based on a bad pun which I implemented in a serious way. The second is a wordplay one, which I threw together on the last day when I thought that other one wasn't going to work. I've been putting together a fun tutorial video on the second one for my two-hit-wonder ghost town youtube channel, though since it contains no cultural icons I'm skeptical it will play well with that audience. I thought maybe I'd upload the video tonight, but now it's getting close to the end of the month, so here is a post to satisfy my weird compulsion, and I will post the video and SIGBOVIK works tomorrow.
Next week I'm going to Paris for the marathon and an art show. Costume ideas? Fun things not to miss?
|Run dressed as the Statue of Liberty. Or the Eiffel Tower?|
|Good ideas, though I guess I should also add that it needs to fit in carry-on luggage and be TSA approved! Then again, how could the Statue of Liberty not be TSA-approved?|
|Nice video! Do you have a source release for that program, or do I have to make it myself? *runs* How does one optimize the output anyways? Obviously, one of the least efficient outputs would be to go in dictionary order and use that chart to link all words by just taking the last letter and going to the start letter of the next word in the dictionary.|
|The website (tom7.org/portmantout/) has a link to the source. Will probably need some minor modifications to get it to build in something other than mingw-x64 on windows.
Aside from the heuristics (choose long suffixes first) and opportunistic improvements (remove words that are incidentally covered), there isn't really any attempt to optimize. It's a good problem (though I'm pretty sure NP-hard). For the video, I ran it in parallel a few hundred thousand times (some parts are arbitrary so I choose randomly) and just saved the smallest one.
|Quick, if you're in the right timezone, you can still post in April!
|Okay, now I'm worried. Are you all right, Tom?
|Thanks for the reminder! I actually forgot because I thought I had already posted this month, so now we have our first cheated back-date!|
|Whew. And I approve of the back-dating.
|Ha ha, good. :) Thanks for reading!|