PERIOD intervals not structured

[lfoppiano]

Given a sentence starting from As early as the 1930s, [...], does it make sense to annotate this chunk as PERIOD

As	B-PERIOD
early	PERIOD
as	PERIOD
the	PERIOD
1930s	PERIOD
,	O

or as it's annotated in the example?

As	O
early	O
as	O
the	O
1930s	B-PERIOD
,	O
Benjamin	B-PERSON
Arditti	PERSON
expressed	O
interest	O
in	O
the	O
history	O
of	O
the	O
Jews	B-PERSON_TYPE
of	O
Bulgaria	B-LOCATION
;	O

[kermitt2]

I would say no because it simply qualifies the period but not changed it.

As early as the 1930s = 1930s

so I would go with the second annotation.

[lfoppiano]

When there is an interval, we annotate from as part of the entity.

What about other elements like

• since: for exampleIt preserves moving pictures since 1919, can be annotated with since 1919 as PERIOD?

• during: related to event can be a period:

  1. during the time of the Nazi occupation:
     • during the time of the <ENAMEX type="EVENT">Nazi occupation</ENAMEX> or
     • <ENAMEX type="PERIOD">during the time of the Nazi occupation</ENAMEX>.
  2. `during the Czarist regime:
     • <ENAMEX type="PERIOD">during the Czarist regime</ENAMEX> or
     • during the <ENAMEX type="LOCATION">Czarist regime</ENAMEX> or
     • during the <ENAMEX type="EVENT">Czarist regime</ENAMEX>

• in: In this example: the staff, chiefs, mayors and their assistants, men who were born in 1919-1924, war prisoners, war refugees from Russia.
  Should in be included in the entity 1919-1924?

[kermitt2]

• since: I would indeed annotate also what qualify the range of period , so I would annotate since, similarly as from, as part of the entity. We don't have the same period type with and without the word since or from

• during: In general I would not annotate during at all, it does not qualify an interval, just "locate" the predicate with respect to the interval

i. any event defines a period, but a period is not an event - so what useful I think is to annotate as EVENT

ii. difficult - I have no clear answer for the moment for this one...

• in: we have a clear interval without the in, it's like during I think, not to be annotated as part of the entity.

[everzeni]

It seems to us that if we annotate Nazi occupation as EVENT, it should be the same class for Czarist regime.

[lfoppiano]

What about after, before?

I have a couple of more cases:

1. after 1930 there was a great boom -> I tend to think that we should annotate after 1930 (same would be for before 1930 blablabla)
2. 7 years after the war there was a great boom -> shall we annotate also after, though I think won't make sense because there is no starting point

Maybe I'm overcomplicating things?

[kermitt2]

@lfoppiano it's not overcomplicating at all, good points to clarify PERIOD !

For me it's exactly like since, from, before, that's defining the period, so I would put both as part of the entity PERIOD : we need these words to translate the raw entity into the correct structured interval of time.

[everzeni]

What about next and last? For example:

The U.S. eventually establishes safe zones west of the Rocky Mountains and spends much of the next decade eradicating zombies in that region.

We feel we should annotate it as PERIOD:

much of the <ENAMEX type="PERIOD">next decade</ENAMEX> eradicating zombies

1. what do you think for next / last ?

2. about the second example of @lfoppiano : 7 years after the war there was a great boom. It seems to me that 7 years after alone makes no sense as a PERIOD... It's different than after 1930. The time marker here is after the war and not only after... maybe we should annotate:

<ENAMEX type="PERIOD">7 years after the war</ENAMEX> there was a great boom

?

[kermitt2]

Thanks @everzeni I agree with both 👍

[everzeni]

question from Slack, should we annotate like 7 years after the war these entities:

1. Seven years after the outbreak began
2. 7 years after the war
   ?

I vote yes, what do you think?

[lfoppiano]

I also vote yes.

Reality doesn't exists ;-) (I'm trying to unbias myself from the fact that it's a fictionary text 🔢 )

[everzeni]

We have another example and we are doubting, I have two questions:

1. about my precedent comment, do you think we should annotate:

<ENAMEX type="PERIOD">Ten years after the official end of the zombie war</ENAMEX>

2. our other case is:

it would have allowed the UK to withhold social benefits to new immigrants for the first four years after they arrived;

Do you think we should annotate:

it would have allowed the <ENAMEX type="LOCATION">UK</ENAMEX> to withhold social 
benefits to new immigrants for the <ENAMEX type="PERIOD">first four years after 
they arrived</ENAMEX>;

thanks!

[everzeni]

Following my previous comment,

We have another sentence and it made us think about the "after X" problem... Here is the sentence:

The Treaties shall cease to apply to the State in question (...) two years after the notification referred to in paragraph 2

How about we decide to annotate inside the PERIOD the part « after X», but only if X is expressing an EVENT? For the recent sentences we discussed, it would lead to:

1. INCLUDE "after X": <ENAMEX type="PERIOD">7 years after the war</ENAMEX>

2. INCLUDE "after X": <ENAMEX type="PERIOD">Ten years after the official end of the zombie war</ENAMEX>

3. EXCLUDE "after X": withhold social benefits to new immigrants for the <ENAMEX type="PERIOD">first four years</ENAMEX> after they arrived

4. EXCLUDE "after X" and annotate MEASURE instead: The Treaties shall cease to apply to the State in question (...) <ENAMEX type="PERIOD">two years</ENAMEX> after the notification referred to in paragraph 2**

What do you think?

[wigdan]

mid afternoon on <ENAMEX type="PERIOD">27 June 2016</ENAMEX>

Would you include (mid after noon) within the whole period?

[lfoppiano]

@everzeni regarding to your first comment, I agree with your proposal. I'm just wondering whether 2 years is a measure or a period. I'm asking because I've seen an example of the same token classified as period in EHRI_unit_il-002798-p_37.2.en.training.xml
--> @lfoppiano I answer you here, yes it's a PERIOD! I edited my comment -- Emilia

@wigdan as you said, I agree not to include mid afternoon on in the whole period.

[kermitt2]

@wigdan @lfoppiano I would add mid afternoon in the annotation, because it is part of the whole time expression:

<ENAMEX type="PERIOD">mid afternoon on 27 June 2016</ENAMEX>

[wigdan]

the pound fell to a seven-year low against the dollar
what would be the appropriate annotation for seven-year?
it indicates time or period, but it comes with (low) which makes it as comparison or measure?
and thats makes me little confused...!

[kermitt2]

I would annotate seven-year as a PERIOD. I don't see any particular impact with low, given that we annotate time "measurement" as PERIOD anyway.

[everzeni]

@kermitt2 what do you think about this comment hereabove: #13 (comment) ? thank you

[kermitt2]

@everzeni it looks convincing ! 👍