checked top-1000 BNC senses in Swe, with some split senses added to Dictionary and DictionaryEng. Wrote bnc-dict-log.txt to describe the procedure, which should be reproducible to other languages now.

lib/src/translator/bnc-dict-log.txt

@@ -0,0 +1,60 @@
+1. Create a check list for Swe
+
+do
+  bnc <- readFile "bnc-to-check.txt" >>= return . words                               -- list of BNC funs
+  dict <- readFile "DictionarySwe.gf" >>= return . map words . lines                  -- current Swe lexicon
+  let dictmap = Data.Map.fromList [(f,unwords ws) | "lin":f:"=":ws <- dict]
+  let bncdict = [(f,maybe "variants{} ;" id $ Data.Map.lookup f dictmap) | f <- bnc]  -- current Swe for BNC
+  writeFile "bncswe.txt" $ unlines [unwords ("lin":f:"=":[ws]) | (f,ws) <- bncdict]   -- print inspectable file
+
+
+2. Inspect the check list
+
+went through one thousand
+- corrected everything
+  - removed trailing comments from corrected entries
+- split senses
+  - added a -- | comment for disambiguating new senses
+
+move the checked words to correctswe.txt
+
+
+3. Apply split senses
+
+grep "\-\- |" correctswe.txt | sort
+
+Copy split senses to bnc-to-check.txt 
+- *but don't remove the unsplit senses* because they are needed to find words from other languages
+
+Copy split senses to Dictionary.gf, together with the -- | comments
+
+Make copies for split senses in DictionaryEng.gf
+
+Verify the result by compiling DictionaryEng.gf 
+
+
+4. Extend the Swe lexicon
+
+do
+  old <- readFile "DictionarySwe.gf" >>= return . map words . lines       -- read old lexicon
+  new <- readFile "correctswe.txt" >>= return . map words . lines         -- read corrected and new words
+  let oldmap = Data.Map.fromList [(f,unwords ws) | "lin":f:"=":ws <- old]
+  let newlist = [(f,unwords (takeWhile (/= "--") ws)) | "lin":f:"=":ws <- new]  -- drop comments from corrected words
+  let newmap = foldr (uncurry Data.Map.insert) oldmap newlist                   -- insert corrected words
+  writeFile "newswe.txt" $ unlines [unwords ("lin":f:"=":[ws]) | (f,ws) <- Data.Map.assocs newmap]  -- print lin rules
+
+Replace the body of DictionarySwe.gf by newswe.txt
+
+Compile DictionarySwe.gf
+
+
+5. Spare the rest to do
+
+Remove the corrected words from bncswe.txt. 
+Or take note of the last word that was checked already.
+
+
+
+
+
+