Skip to content

Commit

Permalink
Added support for understanding Swedish last <day-of-the-week> expres…
Browse files Browse the repository at this point in the history 
…sion

Summary:
One of the most common ways of saying "last \<day-of-the-week\>" in Swedish is saying "i \<day-of-the-week\>s", for example "i tisdags" (last Tuesday) or "i lördags" (last Saturday). This pull request adds support for this.
Closes #92

Reviewed By: blandinw, kodafb

Differential Revision: D6064814

Pulled By: patapizza

fbshipit-source-id: 6ea5466
  • Loading branch information
@facebook-github-bot
Fredrik Wallén authored and facebook-github-bot committed Oct 16, 2017
1 parent 0e95062 commit 1b176c3
Show file tree
Hide file tree
Showing 3 changed files with 52 additions and 26 deletions.
50 changes: 31 additions & 19 deletions Duckling/Ranking/Classifiers/SV_XX.hs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -58,8 +58,8 @@ classifiers
likelihoods = HashMap.fromList [("", 0.0)], n = 1}}),
("Tisdag",
Classifier{okData =
ClassData{prior = 0.0, unseen = -2.3025850929940455,
likelihoods = HashMap.fromList [("", 0.0)], n = 8},
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}}),
Expand Down Expand Up @@ -757,6 +757,16 @@ classifiers
koData =
ClassData{prior = -infinity, unseen = -0.6931471805599453,
likelihoods = HashMap.fromList [], n = 0}}),
("last <day-of-week>",
Classifier{okData =
ClassData{prior = 0.0, unseen = -1.6094379124341003,
likelihoods =
HashMap.fromList
[("Tisdag", -0.6931471805599453), ("day", -0.6931471805599453)],
n = 1},
koData =
ClassData{prior = -infinity, unseen = -1.0986122886681098,
likelihoods = HashMap.fromList [], n = 0}}),
("<time> after next",
Classifier{okData =
ClassData{prior = 0.0, unseen = -2.639057329615259,
Expand Down Expand Up @@ -1587,29 +1597,31 @@ classifiers
likelihoods = HashMap.fromList [], n = 0}}),
("this <time>",
Classifier{okData =
ClassData{prior = -1.276293465905562, unseen = -3.6635616461296463,
ClassData{prior = -1.2992829841302609,
unseen = -3.6888794541139363,
likelihoods =
HashMap.fromList
[("week", -2.9444389791664407),
("intersect", -2.538973871058276),
("season", -2.0281482472922856),
("next <cycle>", -2.9444389791664407),
("day", -2.0281482472922856), ("Oktober", -2.538973871058276),
("hour", -2.2512917986064953), ("month", -2.0281482472922856),
("week-end", -2.2512917986064953)],
[("week", -2.9704144655697013),
("intersect", -2.5649493574615367),
("season", -2.0541237336955462),
("next <cycle>", -2.9704144655697013),
("day", -2.0541237336955462), ("Oktober", -2.5649493574615367),
("hour", -2.277267285009756), ("month", -2.0541237336955462),
("week-end", -2.277267285009756)],
n = 12},
koData =
ClassData{prior = -0.3272129112084162, unseen = -4.343805421853684,
ClassData{prior = -0.3184537311185346, unseen = -4.382026634673881,
likelihoods =
HashMap.fromList
[("September", -2.7212954278522306),
("intersect", -1.845826690498331),
("time-of-day (latent)", -3.2321210516182215),
("Oktober", -1.9328380674879604), ("Mars", -3.2321210516182215),
("noon", -3.6375861597263857), ("hour", -2.7212954278522306),
("month", -0.9985288301111273),
("morning", -3.6375861597263857)],
n = 31}}),
[("Tisdag", -3.676300671907076),
("September", -2.760009940032921),
("intersect", -1.8845412026790211), ("day", -3.676300671907076),
("time-of-day (latent)", -3.270835563798912),
("Oktober", -1.971552579668651), ("Mars", -3.270835563798912),
("noon", -3.676300671907076), ("hour", -2.760009940032921),
("month", -1.0372433422918175),
("morning", -3.676300671907076)],
n = 32}}),
("within <duration>",
Classifier{okData =
ClassData{prior = 0.0, unseen = -1.6094379124341003,
Expand Down
1 change: 1 addition & 0 deletions Duckling/Time/SV/Corpus.hs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -195,6 +195,7 @@ allExamples = concat
]
, examples (datetime (2013, 2, 5, 0, 0, 0) Day)
[ "förra tisdag"
, "i tisdags"
]
, examples (datetime (2013, 2, 19, 0, 0, 0) Day)
[ "nästa tisdag"
Expand Down
27 changes: 20 additions & 7 deletions Duckling/Time/SV/Rules.hs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -29,13 +29,13 @@ import qualified Duckling.TimeGrain.Types as TG

daysOfWeek :: [(Text, String)]
daysOfWeek =
[ ( "Mandag" , "måndag(en)?|mån\\.?" )
, ( "Tisdag" , "tisdag(en)?|tis?\\.?" )
, ( "Onsdag" , "onsdag(en)?|ons\\.?" )
, ( "Torsdag" , "torsdag(en)?|tors?\\.?" )
, ( "Fredag" , "fredag(en)?|fre\\.?" )
, ( "Lordag" , "lördag(en)?|lör\\.?" )
, ( "Sondag" , "söndag(en)?|sön\\.?" )
[ ( "Mandag" , "måndag(en)?s?|mån\\.?" )
, ( "Tisdag" , "tisdag(en)?s?|tis?\\.?" )
, ( "Onsdag" , "onsdag(en)?s?|ons\\.?" )
, ( "Torsdag" , "torsdag(en)?s?|tors?\\.?" )
, ( "Fredag" , "fredag(en)?s?|fre\\.?" )
, ( "Lordag" , "lördag(en)?s?|lör\\.?" )
, ( "Sondag" , "söndag(en)?s?|sön\\.?" )
]

ruleDaysOfWeek :: [Rule]
Expand Down Expand Up @@ -503,6 +503,18 @@ ruleThisnextDayofweek = Rule
_ -> Nothing
}

ruleLastDayofweek :: Rule
ruleLastDayofweek = Rule
{ name = "last <day-of-week>"
, pattern =
[ regex "i"
, Predicate isADayOfWeek
]
, prod = \tokens -> case tokens of
(_:Token Time td:_) -> tt $ predNth (-1) True td
_ -> Nothing
}

ruleTheDayBeforeYesterday :: Rule
ruleTheDayBeforeYesterday = Rule
{ name = "the day before yesterday"
Expand Down Expand Up @@ -1752,6 +1764,7 @@ rules =
, ruleThisPartofday
, ruleThisTime
, ruleThisnextDayofweek
, ruleLastDayofweek
, ruleTimeAfterNext
, ruleTimeBeforeLast
, ruleTimePartofday
Expand Down

0 comments on commit 1b176c3

Please sign in to comment.