[DE DateTimeV2] "dem Jahr" != "this year"

@tellarin Hello Börje 😃

Regarding the issue “aus dem Jahr 2012” returning the current year (for the “dem Jahr” portion):

I have identified line 151 in German-DateTime.yaml for returning the “current time frame” :

ThisRegex: !nestedRegex
  def: (((diese((n|m)|(\s*woche))(\s*am)?\s+){WeekDayRegex})|diese(n|r)?\s*(sommer|winter|frühling|herbst))

• The above expression seems correct in that it contains dieser but not dem
• However, there is no reference to year here. So I dont really know where the dem Jahr behavior is being generated from.

I also found this ⬇️ expression, which contains dem:

RelativeRegex: !simpleRegex
  def: \b(?<order>(über)?nächste[rns]?|kommende[rns]?|diese[rmns]?|vorletzte[snm]?|letzte[rns]?|vergangene[rns]?|vorherige[rns]?|vorige[rns]?|jetzige[rns]?|heutige[rns]?|aktuelle[rns]?|gestrige[rns]?|morgige[rns]?|de[rmsn]|am)

This ⬆️ pattern is referenced by the below ⬇️ expression, maybe it is the cause of the incorrect output:

OneWordPeriodRegex: !nestedRegex
 def: \b((((im\s+)?monat\s+)?({RelativeRegex}\s*(jahr\s*(im\s*)?)?)?(?<month>apr(il|\.)|aug(ust|\.)|dez(ember|\.)|feb(ruar|ber|\.)|j[äa]n(uar|ner|\.)|jul(e?i|l\.)|jun([io]|\.)|märz|mai|nov(ember|\.)|okt(ober|\.)|sept?(ember|\.)))|(?<business>unter\s+der\s+woche)|({RelativeRegex}\s+)?((?<business>werktags|arbeitswoche)|woche(nende)?|monat(s)?|jahr)|(monat|jahr))\b
 references: [ RelativeRegex, AfterNextSuffixRegex ]

---

I have confirmed that this problem is still live on LUIS. This is very high priority for me, it would be great if you could change it soon.

In comparison to English, it is basically the same grammar structure; "the year" != "das Jahr"|"dem Jahr". the year needs specification via a year number to make any sense.

Thanks in advance.

Regards

Brian

[aitelint]

ok, I will try to implement this logic only in German for now, let see what @tellarin says about the English case.

[BrianInGermany]

@LionbridgeCSII

Thanks for the info on the timeframe 😃

But why do you map “the year” to the current year? The only case I can think of for this is “Film of the year”. But this is only for the case “of the year”|“des Jahres”. This does not apply to a sentence like “Show me a film from the year”. This last example is underspecified and the listener would ask “What year exactly?”

Primarily, we want to ensure that “Films from the year 2012” does not return the current year. But if the user says “Show me a film from the year”, we also dont want the current year returned, because this information is not included in the sentence (a different case than “Film of the year”

This above described behavior is what we want for German, but like I said, I personally think it should be the same for English.