Debian Codegraph @

data/ru/tools/README.md

raw
1
2 Regenerating the Russian Dictionaries
3 -------------------------------------
4 May 2014
5
6
7 The raw data is located in the file morphs20030410.mrd. This is the
8 master morphology file from 2003. The corresponding grammatical
9 markup is in rgramtab20030306.tab. The current Russian dicts are built
10 from these two files.
11
12 It is to be converted to Berkley DB files by running:
13 ```
14 time ./make-hash-db.pl
15 ```
16 This will create the files `para.db`, `root.db`, `rule.db` and
17 `tail.db`, as well as several text files: `paradigm-not-used.txt`,
18 `morph-examp.txt` and `ancode-not-defined.txt`. It should take
19 about 45 seconds or less to run. These are then used to create
20 the link-grammar dictionaries:
21 ```
22 time ./mkdict.pl
23 ```
24 This should take about 25 seconds or less. After this, the Russian
25 dictionaries are ready to be used!
26
27 The `query-morph.pl` tool will examine the morphology candidates for a
28 given word. For example:
29 ```
30 ./query-morph.pl ั‚ะตัั‚
31 ```
32
33 Morphology sources
34 ------------------
35 The morphology is nominally a part of http://www.aot.ru It comes from
36 the "seman" project:
37 ```
38 http://seman.sourceforge.net/
39 ```
40 Source data for the morphology is available under the LGPL license at:
41 ```
42 http://sourceforge.net/p/seman/svn/HEAD/tree/trunk/
43 ```
44 A read-only version can be obtained as:
45 ```
46 svn checkout svn://svn.code.sf.net/p/seman/svn/trunk seman-svn
47 ```
48 So, for example, `rgramtab.tab` is in `trunk/Dicts/Morph/` and the
49 `morphs.mrd` is in `trunk/Dicts/SrcMorph/RusSrc/morphs.mrd`. These files
50 are encoded in KOI8-R. The iconv command can be used to convert
51 KOI8-R to UTF, as follows:
52 ```
53 iconv -f KOI8-R -t UTF-8 <infile> > <outfile>
54 ```
55 So:
56 ```
57 iconv -f KOI8-R -t UTF-8 morphs.mrd > morphs-utf8.mrd
58 iconv -f KOI8-R -t UTF-8 rgramtab.tab > rgramtab-utf8.tab
59 ```
60
61 The End.
62 --------