mirror of
https://github.com/facebook/duckling.git
synced 2024-11-29 01:03:44 +03:00
4ef255e577
Summary: generalize chinese digit specifier (十百千万亿) parsing, and add hanzi tests These digits specifiers can be parsed as (<num><speci>)<num>, by using the multiplicater value <num><speci>, and a connect function that adds them together (two cases, skipping digits [which requires a 零 in between], and digits in consecutive locations). Note that 个 is technically a digit specifier, but in Chinese it is never used directly as a numeral specifier, and always as a counter. Reviewed By: zliu41 Differential Revision: D7424249 fbshipit-source-id: 20a85a7df1f908ee9879e92b904178fa26a9a5e5
911 lines
53 KiB
Haskell
911 lines
53 KiB
Haskell
-- Copyright (c) 2016-present, Facebook, Inc.
|
|
-- All rights reserved.
|
|
--
|
|
-- This source code is licensed under the BSD-style license found in the
|
|
-- LICENSE file in the root directory of this source tree. An additional grant
|
|
-- of patent rights can be found in the PATENTS file in the same directory.
|
|
|
|
-----------------------------------------------------------------
|
|
-- Auto-generated by regenClassifiers
|
|
--
|
|
-- DO NOT EDIT UNLESS YOU ARE SURE THAT YOU KNOW WHAT YOU ARE DOING
|
|
-- @generated
|
|
-----------------------------------------------------------------
|
|
{-# LANGUAGE OverloadedStrings #-}
|
|
module Duckling.Ranking.Classifiers.ZH_TW (classifiers) where
|
|
import Data.String
|
|
import Prelude
|
|
import qualified Data.HashMap.Strict as HashMap
|
|
import Duckling.Ranking.Types
|
|
|
|
classifiers :: Classifiers
|
|
classifiers
|
|
= HashMap.fromList
|
|
[("<time> timezone",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -1.6094379124341003,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("<time-of-day> am|pm", -0.6931471805599453),
|
|
("hour", -0.6931471805599453)],
|
|
n = 1},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -1.0986122886681098,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("Thursday",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -1.9459101490553135,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 5},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("integer (numeric)",
|
|
Classifier{okData =
|
|
ClassData{prior = -0.5020919437972361, unseen = -3.871201010907891,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 46},
|
|
koData =
|
|
ClassData{prior = -0.9295359586241757,
|
|
unseen = -3.4657359027997265,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 30}}),
|
|
("the day before yesterday",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -1.3862943611198906,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 2},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("today",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -1.3862943611198906,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 2},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("mm/dd",
|
|
Classifier{okData =
|
|
ClassData{prior = -1.6094379124341003,
|
|
unseen = -1.0986122886681098,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 1},
|
|
koData =
|
|
ClassData{prior = -0.2231435513142097, unseen = -1.791759469228055,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 4}}),
|
|
("absorption of , after named day",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -2.833213344056216,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("day", -0.6931471805599453), ("Sunday", -0.6931471805599453)],
|
|
n = 7},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -1.0986122886681098,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("tonight",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -1.791759469228055,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 4},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("October",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -1.791759469228055,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 4},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("month (grain)",
|
|
Classifier{okData =
|
|
ClassData{prior = -1.0076405104623831,
|
|
unseen = -3.2188758248682006,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 23},
|
|
koData =
|
|
ClassData{prior = -0.45425527227759643,
|
|
unseen = -3.7376696182833684,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 40}}),
|
|
("<time-of-day> o'clock",
|
|
Classifier{okData =
|
|
ClassData{prior = -0.4418327522790392, unseen = -3.044522437723423,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("<integer> (latent time-of-day)", -0.6931471805599453),
|
|
("hour", -0.6931471805599453)],
|
|
n = 9},
|
|
koData =
|
|
ClassData{prior = -1.0296194171811581,
|
|
unseen = -2.5649493574615367,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("<integer> (latent time-of-day)", -0.6931471805599453),
|
|
("hour", -0.6931471805599453)],
|
|
n = 5}}),
|
|
("national day",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -2.3025850929940455,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 8},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("number suffix: \21313",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -3.258096538021482,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 24},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("Wednesday",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -2.9444389791664407,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 17},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("hour (grain)",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -3.0910424533583156,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 20},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("intersect",
|
|
Classifier{okData =
|
|
ClassData{prior = -0.3519764231571781, unseen = -4.983606621708336,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("children's day<part-of-day> <dim time>", -3.367295829986474),
|
|
("year (numeric with year symbol)<named-month> <day-of-month>",
|
|
-1.7578579175523736),
|
|
("dayday", -1.6094379124341003),
|
|
("hourhour", -3.367295829986474),
|
|
("hourminute", -3.367295829986474),
|
|
("absorption of , after named day<named-month> <day-of-month>",
|
|
-1.6094379124341003),
|
|
("dayminute", -3.367295829986474),
|
|
("tonight<time-of-day> o'clock", -3.367295829986474),
|
|
("<dim time> <part-of-day>relative minutes after|past <integer> (hour-of-day)",
|
|
-3.367295829986474),
|
|
("yearday", -1.7578579175523736)],
|
|
n = 64},
|
|
koData =
|
|
ClassData{prior = -1.215022640512521, unseen = -4.276666119016055,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("children's day<part-of-day> <dim time>", -2.065455299705096),
|
|
("dayhour", -2.653241964607215),
|
|
("daymonth", -2.1832383353614793),
|
|
("year (numeric with year symbol)February", -3.164067588373206),
|
|
("<dim time> <part-of-day><time-of-day> o'clock",
|
|
-3.164067588373206),
|
|
("hourhour", -3.164067588373206),
|
|
("year (numeric with year symbol)March", -2.653241964607215),
|
|
("hourminute", -2.653241964607215),
|
|
("yearmonth", -2.316769727986002),
|
|
("dayminute", -2.653241964607215),
|
|
("<dim time> <part-of-day>relative minutes after|past <integer> (hour-of-day)",
|
|
-2.653241964607215),
|
|
("absorption of , after named dayFebruary",
|
|
-2.1832383353614793)],
|
|
n = 27}}),
|
|
("year (grain)",
|
|
Classifier{okData =
|
|
ClassData{prior = -0.8209805520698302,
|
|
unseen = -2.5649493574615367,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 11},
|
|
koData =
|
|
ClassData{prior = -0.579818495252942, unseen = -2.772588722239781,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 14}}),
|
|
("Saturday",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -1.9459101490553135,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 5},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("next <cycle>",
|
|
Classifier{okData =
|
|
ClassData{prior = -0.7472144018302211,
|
|
unseen = -3.2188758248682006,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("week", -1.3862943611198906),
|
|
("month (grain)", -1.791759469228055),
|
|
("year (grain)", -2.4849066497880004),
|
|
("week (grain)", -1.3862943611198906),
|
|
("year", -2.4849066497880004), ("month", -1.791759469228055)],
|
|
n = 9},
|
|
koData =
|
|
ClassData{prior = -0.6418538861723948, unseen = -3.295836866004329,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("week", -0.8602012652231115),
|
|
("week (grain)", -0.8602012652231115)],
|
|
n = 10}}),
|
|
("last year",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -1.3862943611198906,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 2},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("this <day-of-week>",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -3.970291913552122,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("Wednesday", -1.8718021769015913),
|
|
("Monday", -1.8718021769015913), ("day", -0.7323678937132265),
|
|
("Tuesday", -1.5533484457830569)],
|
|
n = 24},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -1.6094379124341003,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("yyyy-mm-dd",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -1.3862943611198906,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 2},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("mm/dd/yyyy",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -1.791759469228055,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 4},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("evening|night",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -1.3862943611198906,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 2},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("Monday",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -3.258096538021482,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 24},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("yesterday",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -1.6094379124341003,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 3},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("next year",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -1.3862943611198906,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 2},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("hh:mm (time-of-day)",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -1.9459101490553135,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 5},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("<integer> (latent time-of-day)",
|
|
Classifier{okData =
|
|
ClassData{prior = -1.4170660197866443, unseen = -2.995732273553991,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("integer (numeric)", -1.3350010667323402),
|
|
("integer (0..10)", -0.3794896217049037)],
|
|
n = 16},
|
|
koData =
|
|
ClassData{prior = -0.2776317365982795,
|
|
unseen = -3.9889840465642745,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("integer (numeric)", -2.3608540011180215),
|
|
("integer (0..10)", -0.28141245943818544),
|
|
("<number>\20010/\20491", -1.890850371872286)],
|
|
n = 50}}),
|
|
("nth <time> of <time>",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -3.332204510175204,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("Octoberordinal (digits)Monday", -1.6863989535702288),
|
|
("monthday", -0.7308875085427924),
|
|
("month (numeric with month symbol)ordinal (digits)Monday",
|
|
-1.0986122886681098)],
|
|
n = 12},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -1.3862943611198906,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("month (numeric with month symbol)",
|
|
Classifier{okData =
|
|
ClassData{prior = -0.10763066419236544,
|
|
unseen = -3.891820298110627,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("integer (numeric)", -1.1631508098056809),
|
|
("number suffix: \21313", -2.2617630984737906),
|
|
("integer (0..10)", -0.5753641449035618)],
|
|
n = 44},
|
|
koData =
|
|
ClassData{prior = -2.2823823856765264,
|
|
unseen = -2.3025850929940455,
|
|
likelihoods =
|
|
HashMap.fromList [("<number>\20010/\20491", -0.40546510810816444)],
|
|
n = 5}}),
|
|
("week (grain)",
|
|
Classifier{okData =
|
|
ClassData{prior = -0.8434293836092833,
|
|
unseen = -3.6635616461296463,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 37},
|
|
koData =
|
|
ClassData{prior = -0.5625269981428811,
|
|
unseen = -3.9318256327243257,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 49}}),
|
|
("valentine's day",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -1.3862943611198906,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 2},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("year (numeric with year symbol)",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -2.772588722239781,
|
|
likelihoods = HashMap.fromList [("integer (numeric)", 0.0)],
|
|
n = 14},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("now",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -2.4849066497880004,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 10},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("numbers prefix with -, negative or minus",
|
|
Classifier{okData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0},
|
|
koData =
|
|
ClassData{prior = 0.0, unseen = -1.791759469228055,
|
|
likelihoods = HashMap.fromList [("integer (numeric)", 0.0)],
|
|
n = 4}}),
|
|
("Friday",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -1.9459101490553135,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 5},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("tomorrow",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -1.6094379124341003,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 3},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("this year",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -1.0986122886681098,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 1},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("next <day-of-week>",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -3.1780538303479458,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("Wednesday", -1.3437347467010947),
|
|
("day", -0.7375989431307791), ("Tuesday", -1.3437347467010947)],
|
|
n = 10},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -1.3862943611198906,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("fractional number",
|
|
Classifier{okData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0},
|
|
koData =
|
|
ClassData{prior = 0.0, unseen = -1.9459101490553135,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 5}}),
|
|
("Sunday",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -3.0910424533583156,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 20},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("afternoon",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -2.3978952727983707,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 9},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("<duration> from now",
|
|
Classifier{okData =
|
|
ClassData{prior = -infinity, unseen = -2.1972245773362196,
|
|
likelihoods = HashMap.fromList [], n = 0},
|
|
koData =
|
|
ClassData{prior = 0.0, unseen = -3.891820298110627,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("week", -2.2617630984737906), ("second", -2.772588722239781),
|
|
("day", -2.2617630984737906), ("year", -2.772588722239781),
|
|
("<integer> <unit-of-duration>", -0.8266785731844679),
|
|
("hour", -2.2617630984737906), ("month", -2.772588722239781),
|
|
("minute", -2.772588722239781)],
|
|
n = 20}}),
|
|
("February",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -2.833213344056216,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 15},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("this <cycle>",
|
|
Classifier{okData =
|
|
ClassData{prior = -1.0185695809945732,
|
|
unseen = -3.4339872044851463,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("week", -0.916290731874155),
|
|
("year (grain)", -2.3025850929940455),
|
|
("week (grain)", -0.916290731874155),
|
|
("year", -2.3025850929940455)],
|
|
n = 13},
|
|
koData =
|
|
ClassData{prior = -0.4480247225269604,
|
|
unseen = -3.9318256327243257,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("week", -0.7339691750802004),
|
|
("week (grain)", -0.7339691750802004)],
|
|
n = 23}}),
|
|
("minute (grain)",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -2.890371757896165,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 16},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("<dim time> <part-of-day>",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -4.248495242049359,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("dayhour", -0.7375989431307791),
|
|
("national dayevening|night", -3.1354942159291497),
|
|
("<named-month> <day-of-month>morning", -1.1895840668738362),
|
|
("children's dayafternoon", -3.1354942159291497),
|
|
("Mondaymorning", -2.03688192726104)],
|
|
n = 32},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -1.791759469228055,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("<part-of-day> <dim time>",
|
|
Classifier{okData =
|
|
ClassData{prior = -0.6097655716208943, unseen = -4.07753744390572,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("tonight<integer> (latent time-of-day)", -2.451005098112319),
|
|
("hourhour", -1.3523928094442093),
|
|
("afternoonrelative minutes after|past <integer> (hour-of-day)",
|
|
-1.6625477377480489),
|
|
("afternoon<time-of-day> o'clock", -2.6741486494265287),
|
|
("hourminute", -1.575536360758419),
|
|
("afternoon<integer> (latent time-of-day)",
|
|
-2.6741486494265287),
|
|
("afternoonhh:mm (time-of-day)", -3.367295829986474),
|
|
("tonight<time-of-day> o'clock", -2.451005098112319)],
|
|
n = 25},
|
|
koData =
|
|
ClassData{prior = -0.7841189587656721,
|
|
unseen = -3.9318256327243257,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("hourhour", -1.4271163556401458),
|
|
("afternoonrelative minutes after|past <integer> (hour-of-day)",
|
|
-1.5141277326297755),
|
|
("afternoon<time-of-day> o'clock", -2.120263536200091),
|
|
("hourminute", -1.5141277326297755),
|
|
("afternoon<integer> (latent time-of-day)",
|
|
-1.9661128563728327)],
|
|
n = 21}}),
|
|
("<integer> <unit-of-duration>",
|
|
Classifier{okData =
|
|
ClassData{prior = -infinity, unseen = -3.044522437723423,
|
|
likelihoods = HashMap.fromList [], n = 0},
|
|
koData =
|
|
ClassData{prior = 0.0, unseen = -5.905361848054571,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("week", -2.6837575085331653),
|
|
("integer (0..10)month (grain)", -2.535337503414892),
|
|
("integer (0..10)hour (grain)", -3.06941998934515),
|
|
("<number>\20010/\20491week (grain)", -3.5047380606029956),
|
|
("second", -3.2635760037861075),
|
|
("integer (0..10)day (grain)", -3.06941998934515),
|
|
("number suffix: \21313month (grain)", -4.293195420967266),
|
|
("integer (0..10)year (grain)", -3.6000482404073204),
|
|
("<number>\20010/\20491month (grain)", -3.130044611161585),
|
|
("integer (numeric)year (grain)", -3.1945831322991562),
|
|
("integer (0..10)second (grain)", -3.2635760037861075),
|
|
("day", -3.06941998934515), ("year", -2.7245795030534206),
|
|
("integer (0..10)minute (grain)", -3.06941998934515),
|
|
("hour", -2.8581108956779433),
|
|
("integer (0..10)week (grain)", -3.1945831322991562),
|
|
("month", -1.7754989483562746),
|
|
("<number>\20010/\20491hour (grain)", -4.293195420967266),
|
|
("integer (numeric)month (grain)", -3.1945831322991562),
|
|
("minute", -3.06941998934515)],
|
|
n = 173}}),
|
|
("<time-of-day> am|pm",
|
|
Classifier{okData =
|
|
ClassData{prior = -0.4700036292457356, unseen = -2.70805020110221,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("hh:mm (time-of-day)", -1.252762968495368),
|
|
("<integer> (latent time-of-day)", -1.540445040947149),
|
|
("hour", -1.540445040947149), ("minute", -1.252762968495368)],
|
|
n = 5},
|
|
koData =
|
|
ClassData{prior = -0.9808292530117262,
|
|
unseen = -2.3978952727983707,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("<integer> (latent time-of-day)", -0.916290731874155),
|
|
("hour", -0.916290731874155)],
|
|
n = 3}}),
|
|
("relative minutes after|past <integer> (hour-of-day)",
|
|
Classifier{okData =
|
|
ClassData{prior = -0.6931471805599453,
|
|
unseen = -3.2188758248682006,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("<integer> (latent time-of-day)integer with consecutive unit modifiers",
|
|
-0.7801585575495751),
|
|
("hour", -0.7801585575495751)],
|
|
n = 10},
|
|
koData =
|
|
ClassData{prior = -0.6931471805599453,
|
|
unseen = -3.2188758248682006,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("<integer> (latent time-of-day)number suffix: \21313",
|
|
-1.3862943611198906),
|
|
("<integer> (latent time-of-day)integer (0..10)",
|
|
-1.3862943611198906),
|
|
("hour", -0.7801585575495751)],
|
|
n = 10}}),
|
|
("army's day",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -1.3862943611198906,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 2},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("intersect by \",\"",
|
|
Classifier{okData =
|
|
ClassData{prior = -0.2231435513142097, unseen = -4.110873864173311,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("Sunday<named-month> <day-of-month>", -0.7270487322356266),
|
|
("dayday", -0.7270487322356266)],
|
|
n = 28},
|
|
koData =
|
|
ClassData{prior = -1.6094379124341003,
|
|
unseen = -2.9444389791664407,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("daymonth", -0.8109302162163288),
|
|
("SundayFebruary", -0.8109302162163288)],
|
|
n = 7}}),
|
|
("integer with consecutive unit modifiers",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -3.367295829986474,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("number suffix: \21313integer (0..10)", -0.6931471805599453),
|
|
("integer (0..10)integer (0..10)", -0.6931471805599453)],
|
|
n = 26},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -1.0986122886681098,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("second (grain)",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -2.70805020110221,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 13},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("<duration> ago",
|
|
Classifier{okData =
|
|
ClassData{prior = -infinity, unseen = -2.1972245773362196,
|
|
likelihoods = HashMap.fromList [], n = 0},
|
|
koData =
|
|
ClassData{prior = 0.0, unseen = -3.891820298110627,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("week", -2.2617630984737906), ("second", -2.772588722239781),
|
|
("day", -2.2617630984737906), ("year", -2.772588722239781),
|
|
("<integer> <unit-of-duration>", -0.8266785731844679),
|
|
("hour", -2.2617630984737906), ("month", -2.772588722239781),
|
|
("minute", -2.772588722239781)],
|
|
n = 20}}),
|
|
("last <time>",
|
|
Classifier{okData =
|
|
ClassData{prior = -1.2286654169163076,
|
|
unseen = -3.4657359027997265,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("day", -0.8690378470236094), ("Sunday", -1.3545456628053103),
|
|
("Tuesday", -1.6422277352570913)],
|
|
n = 12},
|
|
koData =
|
|
ClassData{prior = -0.3462762367178338, unseen = -4.189654742026425,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("<integer> (latent time-of-day)", -0.916290731874155),
|
|
("month (numeric with month symbol)", -2.5649493574615367),
|
|
("hour", -0.916290731874155), ("month", -2.5649493574615367)],
|
|
n = 29}}),
|
|
("March",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -2.1972245773362196,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 7},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("the day after tomorrow",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -1.6094379124341003,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 3},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("next <time>",
|
|
Classifier{okData =
|
|
ClassData{prior = -1.6486586255873816, unseen = -2.890371757896165,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("day", -1.041453874828161), ("Tuesday", -1.041453874828161)],
|
|
n = 5},
|
|
koData =
|
|
ClassData{prior = -0.21357410029805904,
|
|
unseen = -3.912023005428146,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("Wednesday", -2.1000608288825715),
|
|
("<integer> (latent time-of-day)", -1.1192315758708455),
|
|
("month (numeric with month symbol)", -3.1986731175506815),
|
|
("day", -2.1000608288825715), ("hour", -1.1192315758708455),
|
|
("month", -3.1986731175506815)],
|
|
n = 21}}),
|
|
("last <cycle>",
|
|
Classifier{okData =
|
|
ClassData{prior = -0.8472978603872037,
|
|
unseen = -3.2188758248682006,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("week", -1.3862943611198906),
|
|
("month (grain)", -1.791759469228055),
|
|
("year (grain)", -2.4849066497880004),
|
|
("week (grain)", -1.3862943611198906),
|
|
("year", -2.4849066497880004), ("month", -1.791759469228055)],
|
|
n = 9},
|
|
koData =
|
|
ClassData{prior = -0.5596157879354228,
|
|
unseen = -3.4339872044851463,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("week", -0.8362480242006186),
|
|
("week (grain)", -0.8362480242006186)],
|
|
n = 12}}),
|
|
("christmas",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -1.791759469228055,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 4},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("new year's day",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -1.6094379124341003,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 3},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("next n <cycle>",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -4.61512051684126,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("week", -2.4079456086518722),
|
|
("integer (0..10)hour (grain)", -2.659260036932778),
|
|
("<number>\20010/\20491week (grain)", -2.995732273553991),
|
|
("second", -2.8134107167600364),
|
|
("integer (0..10)day (grain)", -2.5257286443082556),
|
|
("integer (0..10)year (grain)", -3.2188758248682006),
|
|
("<number>\20010/\20491month (grain)", -2.8134107167600364),
|
|
("integer (0..10)second (grain)", -2.8134107167600364),
|
|
("day", -2.5257286443082556), ("year", -3.2188758248682006),
|
|
("integer (0..10)minute (grain)", -2.659260036932778),
|
|
("hour", -2.4079456086518722),
|
|
("integer (0..10)week (grain)", -2.995732273553991),
|
|
("month", -2.8134107167600364),
|
|
("<number>\20010/\20491hour (grain)", -3.506557897319982),
|
|
("minute", -2.659260036932778)],
|
|
n = 42},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -2.833213344056216,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("Tuesday",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -3.295836866004329,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 25},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("children's day",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -1.791759469228055,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 4},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("<named-month> <day-of-month>",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -5.231108616854587,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("month (numeric with month symbol)integer with consecutive unit modifiers",
|
|
-2.392533329656985),
|
|
("Marchinteger (0..10)", -3.4339872044851463),
|
|
("Marchinteger (numeric)", -4.127134385045092),
|
|
("month (numeric with month symbol)integer (numeric)",
|
|
-2.517696472610991),
|
|
("Februaryinteger (0..10)", -3.146305132033365),
|
|
("month (numeric with month symbol)number suffix: \21313",
|
|
-3.146305132033365),
|
|
("month", -0.7371103039810614),
|
|
("month (numeric with month symbol)integer (0..10)",
|
|
-2.517696472610991),
|
|
("Februaryinteger with consecutive unit modifiers",
|
|
-2.392533329656985),
|
|
("Februarynumber suffix: \21313", -3.146305132033365)],
|
|
n = 88},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -2.3978952727983707,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("integer (0..10)",
|
|
Classifier{okData =
|
|
ClassData{prior = -0.38372512147175847,
|
|
unseen = -5.236441962829949,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 186},
|
|
koData =
|
|
ClassData{prior = -1.143563676530376, unseen = -4.48863636973214,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 87}}),
|
|
("last n <cycle>",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -4.605170185988091,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("week", -2.9856819377004897),
|
|
("integer (0..10)month (grain)", -3.4965075614664802),
|
|
("integer (0..10)hour (grain)", -2.3978952727983707),
|
|
("second", -2.649209701079277),
|
|
("integer (0..10)day (grain)", -2.9856819377004897),
|
|
("integer (0..10)year (grain)", -3.4965075614664802),
|
|
("<number>\20010/\20491month (grain)", -2.3978952727983707),
|
|
("integer (0..10)second (grain)", -2.649209701079277),
|
|
("day", -2.9856819377004897), ("year", -3.4965075614664802),
|
|
("integer (0..10)minute (grain)", -2.3978952727983707),
|
|
("hour", -2.3978952727983707),
|
|
("integer (0..10)week (grain)", -2.9856819377004897),
|
|
("month", -2.1972245773362196),
|
|
("minute", -2.3978952727983707)],
|
|
n = 42},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -2.772588722239781,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("ordinal (digits)",
|
|
Classifier{okData =
|
|
ClassData{prior = -0.6931471805599453,
|
|
unseen = -1.9459101490553135,
|
|
likelihoods =
|
|
HashMap.fromList [("<number>\20010/\20491", -0.1823215567939546)],
|
|
n = 4},
|
|
koData =
|
|
ClassData{prior = -0.6931471805599453,
|
|
unseen = -1.9459101490553135,
|
|
likelihoods =
|
|
HashMap.fromList [("integer (0..10)", -0.1823215567939546)],
|
|
n = 4}}),
|
|
("n <cycle> last",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -4.04305126783455,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("week", -2.4159137783010487),
|
|
("integer (0..10)hour (grain)", -2.9267394020670396),
|
|
("<number>\20010/\20491week (grain)", -2.9267394020670396),
|
|
("second", -2.9267394020670396),
|
|
("integer (0..10)day (grain)", -2.4159137783010487),
|
|
("integer (0..10)year (grain)", -2.9267394020670396),
|
|
("<number>\20010/\20491month (grain)", -2.9267394020670396),
|
|
("integer (0..10)second (grain)", -2.9267394020670396),
|
|
("day", -2.4159137783010487), ("year", -2.9267394020670396),
|
|
("integer (0..10)minute (grain)", -2.9267394020670396),
|
|
("hour", -2.4159137783010487),
|
|
("integer (0..10)week (grain)", -2.9267394020670396),
|
|
("month", -2.9267394020670396),
|
|
("<number>\20010/\20491hour (grain)", -2.9267394020670396),
|
|
("minute", -2.9267394020670396)],
|
|
n = 20},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -2.833213344056216,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("morning",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -2.890371757896165,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 16},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("week-end",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -1.3862943611198906,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 2},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("day (grain)",
|
|
Classifier{okData =
|
|
ClassData{prior = -6.453852113757118e-2,
|
|
unseen = -2.833213344056216,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 15},
|
|
koData =
|
|
ClassData{prior = -2.772588722239781, unseen = -1.0986122886681098,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 1}}),
|
|
("labor day",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -1.3862943611198906,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 2},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("women's day",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -1.3862943611198906,
|
|
likelihoods = HashMap.fromList [("", 0.0)], n = 2},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("<number>\20010/\20491",
|
|
Classifier{okData =
|
|
ClassData{prior = 0.0, unseen = -3.4339872044851463,
|
|
likelihoods = HashMap.fromList [("integer (0..10)", 0.0)], n = 29},
|
|
koData =
|
|
ClassData{prior = -infinity, unseen = -0.6931471805599453,
|
|
likelihoods = HashMap.fromList [], n = 0}}),
|
|
("this <time>",
|
|
Classifier{okData =
|
|
ClassData{prior = -0.2076393647782445, unseen = -4.0943445622221,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("Wednesday", -1.9980959022258835),
|
|
("Monday", -1.9980959022258835), ("day", -0.8586616190375187),
|
|
("hour", -2.9789251552376097), ("Tuesday", -1.6796421711073488),
|
|
("week-end", -2.9789251552376097)],
|
|
n = 26},
|
|
koData =
|
|
ClassData{prior = -1.6739764335716716, unseen = -2.995732273553991,
|
|
likelihoods =
|
|
HashMap.fromList
|
|
[("<integer> (latent time-of-day)", -0.9985288301111273),
|
|
("hour", -0.9985288301111273)],
|
|
n = 6}})] |