|
Last change
on this file since 33166 was 33166, checked in by davidb, 5 years ago |
|
Collection config files and initial programming work for atea collections
|
|
File size:
1.3 KB
|
| Line | |
|---|
| 1 | [
|
|---|
| 2 | [
|
|---|
| 3 | "language_code",
|
|---|
| 4 | "ru"
|
|---|
| 5 | ],
|
|---|
| 6 | [
|
|---|
| 7 | "format_version",
|
|---|
| 8 | "2.4"
|
|---|
| 9 | ],
|
|---|
| 10 | [
|
|---|
| 11 | "pymorphy2_version",
|
|---|
| 12 | "0.8"
|
|---|
| 13 | ],
|
|---|
| 14 | [
|
|---|
| 15 | "compiled_at",
|
|---|
| 16 | "2015-03-15T21:59:48.477191"
|
|---|
| 17 | ],
|
|---|
| 18 | [
|
|---|
| 19 | "source",
|
|---|
| 20 | "opencorpora.org"
|
|---|
| 21 | ],
|
|---|
| 22 | [
|
|---|
| 23 | "source_version",
|
|---|
| 24 | "0.92"
|
|---|
| 25 | ],
|
|---|
| 26 | [
|
|---|
| 27 | "source_revision",
|
|---|
| 28 | "393658"
|
|---|
| 29 | ],
|
|---|
| 30 | [
|
|---|
| 31 | "source_lexemes_count",
|
|---|
| 32 | 389835
|
|---|
| 33 | ],
|
|---|
| 34 | [
|
|---|
| 35 | "source_links_count",
|
|---|
| 36 | 256468
|
|---|
| 37 | ],
|
|---|
| 38 | [
|
|---|
| 39 | "gramtab_length",
|
|---|
| 40 | 4753
|
|---|
| 41 | ],
|
|---|
| 42 | [
|
|---|
| 43 | "gramtab_formats",
|
|---|
| 44 | {
|
|---|
| 45 | "opencorpora-int": "gramtab-opencorpora-int.json",
|
|---|
| 46 | "opencorpora-ext": "gramtab-opencorpora-ext.json"
|
|---|
| 47 | }
|
|---|
| 48 | ],
|
|---|
| 49 | [
|
|---|
| 50 | "paradigms_length",
|
|---|
| 51 | 3163
|
|---|
| 52 | ],
|
|---|
| 53 | [
|
|---|
| 54 | "suffixes_length",
|
|---|
| 55 | 15463
|
|---|
| 56 | ],
|
|---|
| 57 | [
|
|---|
| 58 | "words_dawg_length",
|
|---|
| 59 | 5096128
|
|---|
| 60 | ],
|
|---|
| 61 | [
|
|---|
| 62 | "compile_options",
|
|---|
| 63 | {
|
|---|
| 64 | "max_suffix_length": 5,
|
|---|
| 65 | "paradigm_prefixes": [
|
|---|
| 66 | "",
|
|---|
| 67 | "пП",
|
|---|
| 68 | "МаО"
|
|---|
| 69 | ],
|
|---|
| 70 | "min_ending_freq": 2,
|
|---|
| 71 | "min_paradigm_popularity": 3
|
|---|
| 72 | }
|
|---|
| 73 | ],
|
|---|
| 74 | [
|
|---|
| 75 | "prediction_suffixes_dawg_lengths",
|
|---|
| 76 | [
|
|---|
| 77 | 366134,
|
|---|
| 78 | 1929,
|
|---|
| 79 | 21
|
|---|
| 80 | ]
|
|---|
| 81 | ],
|
|---|
| 82 | [
|
|---|
| 83 | "P(t|w)",
|
|---|
| 84 | true
|
|---|
| 85 | ],
|
|---|
| 86 | [
|
|---|
| 87 | "P(t|w)_unique_words",
|
|---|
| 88 | 21121
|
|---|
| 89 | ],
|
|---|
| 90 | [
|
|---|
| 91 | "P(t|w)_outcomes",
|
|---|
| 92 | 248127
|
|---|
| 93 | ],
|
|---|
| 94 | [
|
|---|
| 95 | "P(t|w)_min_word_freq",
|
|---|
| 96 | 1
|
|---|
| 97 | ],
|
|---|
| 98 | [
|
|---|
| 99 | "corpus_revision",
|
|---|
| 100 | "3725883"
|
|---|
| 101 | ]
|
|---|
| 102 | ] |
|---|
Note:
See
TracBrowser
for help on using the repository browser.
