Commit 240802de0b93846a8e777599b8263edde941e462
1 parent
5b9fa813
test listy frekwencyjnej NKJP1M
Showing
3 changed files
with
1671 additions
and
2532 deletions
Too many changes to show.
To preserve performance only 2 of 3 files are displayed.
corpora/freq_test.ml
... | ... | @@ -2,9 +2,10 @@ |
2 | 2 | open Xstd |
3 | 3 | |
4 | 4 | let _ = |
5 | - let l = File.load_tab "../resources/NKJP1M/NKJP1M_frequencies.tab" (function | |
5 | + let l = File.load_tab "../resources/NKJP1M/NKJP1M-frequency.tab" (function | |
6 | 6 | [orth; lemma; interp; freq] -> orth, lemma, interp, int_of_string freq |
7 | - | _ -> failwith "load_frequencies") in | |
7 | + | l -> failwith ("load_frequencies: " ^ String.concat "\t" l)) in | |
8 | 8 | let qmap = Xlist.fold l StringQMap.empty (fun qmap (orth, lemma, interp, freq) -> |
9 | + let interp = List.hd (Xstring.split ":" interp) in | |
9 | 10 | StringQMap.add_val qmap (lemma ^ "\t" ^ interp) freq) in |
10 | 11 | StringQMap.iter qmap (fun k v -> Printf.printf "%d\t%s\n" v k) |
... | ... |
parser/pipe.ml
... | ... | @@ -188,9 +188,6 @@ let _ = |
188 | 188 | - dopasowanie do siebie tokenów w różnych wersjach tokenizacji |
189 | 189 | - dopasowanie do siebie akapitów i zdań |
190 | 190 | - uporządkowanie etykiet: labels, last itp. |
191 | - 05.10.2016 | |
192 | - - poprawienie uruchamiania mate parsera | |
193 | - - rysowanie struktury zależnościowej generowanejprzesz eniama | |
194 | 191 | *) |
195 | 192 | |
196 | 193 | |
... | ... |