[Introduction] Big5fix will try to fix BIG5 text which was corrupted when transmitted through non-8bits clean(non-iso8859-1) network environment. The text fix rate is not satisfised but can be acceptable with averagely 95% for Chinese text 80% for English text and totally at 90% for overall txt. [installation] change FIXDBDIR in Makefile for word tables make depend (ignore any errors) make make install [Usage] 1. show help: big5fix -h verbose message: big5fix -v change libdir: big5fix -d libdir 2. fix corrupted text from standard input: cat corrupt.sample | big5fix 3. fix corrupted text from command argument: big5fix corrupt.sample 4. examine fix rate if you have the original non-corrupted text, big5fix corrupt.sample > corrupt.fix big5cmp corrupt.txt corrupt.fix [TODO] 1. The structure of word table is not efficient. 2. Should parse the sentence semantically when ambiguity met. Please return any comments to me at skhuang@csie.nctu.edu.tw