Re: [Haskell-cafe] Re: ANN: bitspeak 0.0.1
On Mon, 21 Jun 2010, Maurício CA wrote: bitspeak is a small proof of concept application that allows writing text using only two commands (yes/no, 1/2, top/down etc.). Looks cool! Did you forget any dependencies tho? I get the following error: Oops... Three modules ended up missing in .cabal file. Just uploaded 0.0.2 to hackage, it should work. Two this end, before uploading a package to Hackage I run a script in order to check whether the package can build: $ cat cabal-test package=`basename $1 .tar.gz` tar xfz ./dist/$package.tar.gz --directory=/tmp/ cd /tmp/$package/ runhaskell Setup configure --user runhaskell Setup build runhaskell Setup haddock echo echo "'cabal check' says" cabal check echo echo "After running tests you may want to call:" echo rm -r /tmp/$package/ Run it like $ cabal-test dist/bitspeak-0.0.1.tar.gz___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
[Haskell-cafe] Re: ANN: bitspeak 0.0.1
Sure, Huffman was actually my first tought. But I couldn't think of a pratical display for the result of Huffman encoding that could be easily followed by a human looking at the screen. Since it's an optimal code, letters would not be grouped in alphabetical order. There is a compromise. There is such a thing as an ORDERED Huffman code. Consider a set of strings. If they call begin with the same first letter, assume that letter and consider the suffixes instead. Otherwise, choose a letter L such that as close as possible to half of the strings begin with a letter preceding L in the alphabet as close as possible to half of the strings begin with the letter L or a later letter. I believe that's what I've done. I use this file to give weight to letters: http://bitbucket.org/mauricio/bitspeak/src/tip/src/Corpora.hs After each letter is selected I keep only sufixes after that letter, and then I use 'splitData' to find the point where I'll separate both halves: http://bitbucket.org/mauricio/bitspeak/src/tip/src/Main.hs Best, Maurício ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Re: ANN: bitspeak 0.0.1
On Jun 22, 2010, at 1:26 PM, Maurí cio CA wrote: Sure, Huffman was actually my first tought. But I couldn't think of a pratical display for the result of Huffman encoding that could be easily followed by a human looking at the screen. Since it's an optimal code, letters would not be grouped in alphabetical order. There is a compromise. There is such a thing as an ORDERED Huffman code. Consider a set of strings. If they call begin with the same first letter, assume that letter and consider the suffixes instead. Otherwise, choose a letter L such that as close as possible to half of the strings begin with a letter preceding L in the alphabet as close as possible to half of the strings begin with the letter L or a later letter. ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Re: ANN: bitspeak 0.0.1
On Mon, 21 Jun 2010, Maurício CA wrote: > > > bitspeak is a small proof of concept application that allows > > > writing text using only two commands (yes/no, 1/2, top/down etc.). > > > > There is a parallel between data compression algorithms and this sort of > > task, expressing a sentence in the minimal number of bits via > > compression also minimized the number of yes/no questions that need to > > be asked. > > > > In particular, a Huffman coding: > > Sure, Huffman was actually my first tought. But I couldn't think > of a pratical display for the result of Huffman encoding that > could be easily followed by a human looking at the screen. http://www.inference.phy.cam.ac.uk/dasher/ Tony. -- f.anthony.n.finchhttp://dotat.at/ SOUTH BISCAY: EASTERLY OR NORTHEASTERLY 4 OR 5, OCCASIONALLY 6 IN SOUTHWEST. SLIGHT OR MODERATE. FAIR. GOOD.___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
[Haskell-cafe] Re: ANN: bitspeak 0.0.1
bitspeak is a small proof of concept application that allows writing text using only two commands (yes/no, 1/2, top/down etc.). There is a parallel between data compression algorithms and this sort of task, expressing a sentence in the minimal number of bits via compression also minimized the number of yes/no questions that need to be asked. In particular, a Huffman coding: Sure, Huffman was actually my first tought. But I couldn't think of a pratical display for the result of Huffman encoding that could be easily followed by a human looking at the screen. Since it's an optimal code, letters would not be grouped in alphabetical order. Thinking again, this could be easily accomplished... I could just list the alphabet and the next bit to be choosed below each letter. TODO for 0.1. Thanks! Maurício ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
[Haskell-cafe] Re: ANN: bitspeak 0.0.1
bitspeak is a small proof of concept application that allows writing text using only two commands (yes/no, 1/2, top/down etc.). Looks cool! Did you forget any dependencies tho? I get the following error: Oops... Three modules ended up missing in .cabal file. Just uploaded 0.0.2 to hackage, it should work. Thanks! Maurício ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe