National Gallery Display Date Descriptors

Display date descriptions from TMS:

This page is intended to describe possible search-able semantic definitions of the dates found in the TMS DisplayDate field for the paintings in The National Gallery. Any comments or suggestions would be very helpful.

Part of the National_Gallery_API discussion.

• Semantic Date Descriptors for Artists
• Semantic Date Descriptors for Paintings
• Semantic Date Descriptors Perl Script

The main question to be answered is:

Given the language used within the DisplayDate field; which paintings/artists do we want someone to find when they run a date based search for a particular year or even a range of years?

The intention in the semantic description of the Paintings/Objects and Artists is to define property links to the date field and to establish the beginning and end of the time-span being defined by the DisplayDate. The information contained within the field will be split up into:

Proposed simplifications of the display date field

These adjustments are currently being used to simplify and standardize the date field to make it easier to extrapolate the beginning and end of specific time-spans. They are not currently being used to change the displaydate field, but some of the adjustments may be useful for this in the future.

This set of rules below can be adjusted they are simply a proposed set that seem to make sense.

• Standardize the use of non alphanumeric characters: -,/();
• Standardize how prefixes are displayed: 'Mid 1560s
• Remove unneeded terms, from.
• Ensure all century related time-spans use numbers, 15th Century rather than Fifteenth Century.
• Simplify the range of descriptors that are used than tend to mean the same thing: (The terms used and discarded here can be updated and adjusted as required).
  • All probably, perhaps and ? prefixes have been removed from the date. These terms relate to a degree of confidence.
  • All between prefixes, eg.between XXXX and YYYY have been replaced with XXXX-YYYY.
  • All around prefixes, eg.around XXXX have been replaced with About XXXX.
  • All references to circa have been replaced with About XXXX.
  • All towards, not later than', earlier, about earlier and earlier about prefixes, have been replaced with Before XXXX. This may need to be adjusted if the definitions of Before are changed.
  • The use of the words shortly and soon need to be defined, they are currently removed.
• The word later as in later XXth century has been replaced with late, eg late XXth century.
• Any display date that presents two possible dates, eg. late XXXXs or XXXXs will be split into 2 probable/possible time-spans for the same event.
• All words relating to the actual event, born, died, active, etc are moved out of the time-span.

What information is being extracted from the DisplayDate field ?

• Degree of Confidence: "with time span", "probably within time span" and "possibly within time span". Others can be added but these three seem to cover everything so far.
• The Event: born, died, completed, produced, active etc. What event does the described time span refer.
• The Time-Span: an actual period of time, days, months, years, decades, centuries, etc.

Degrees of Confidence: Example Time-Span properties

• P4F.has_time-span:- Secure date definition.
• PN4F.has_probable_time_span:- Probable date, ("probably", "perhaps" and "?" prefixes).
• PN3F.has_possable_time_span:- Possible date, (Generally used when more than one possible time span has been presented, eg. 1758 or 1764).

The Event: What are we actually defining a time-span for

The following listed included all of the events currently defined or

Paintings/Objects

• Production:
• Start of Production:
• Completion of Production:
• Exhibited:
• Reworked:
• Cast:
• Modelled:

Artists

• Lived:
• Born:
• Died:
• Active:
• Documented:

Defining Time-Span Ranges

The examples listed below are limited to those actual found in the current DisplayDate fields. Other can be added if required.

Simple years and ranges

• XXXX: Basic year descriptor: XXXX-01-01 -> XXXX-12-31
• XXXX-Y: Simple year range from XXXX-01-01 -> XXXY-12-31
• XXXX-YY: Simple year range from XXXX-01-01 -> XXYY-12-31
• XXXX-YYYY: Simple year range from XXXX-01-01 -> YYYY-12-31
• XXXXs: Decade, range from XXX0-01-01 to XXX9-12-31
• ABth century: Century, range from AC00-01-01 -> AC99-12-31, where C = B - 1.
• ABth-DEth century: Multiple Centuries, range from AC00-01-01 -> DF99-12-31, where C = B - 1 and F = E - 1.

The 'About' Prefix

• About XXXX: XXXX(+/-2)
• About XXXX-Y: XXXX(-2) -> XXXY(+2)
• About XXXX-YY: XXXX(-2) -> XXYY(+2)
• About XXXX-YYYY: XXXX(-2) -> YYYY(+2)
• About XXXXs: XXX0 -> XXX9(+5)
• About XXXX-YYYYs: XXXX(-2) -> YYY9(+5)
• XXXX-about YYYY: XXXX -> YYYY(+2)

After and Before

These are difficult terms to define. We can reasonably assume that we would not want someone searching for the year 1950 to find a painting produced After 1640, but searching for which years should find this painting ? These time-spans can these be described as an open range, with only a beginning or end, but this, as the above example shows, could be problematic.

Simple

• After XXXX: XXXX(+1) -> XXXX(+5)
• Before XXXX: XXXX(-5) -> XXXX(-1)

Ranges

After or Before a range gets slightly more complicated. Can these be described again as an open range, with only a beginning or end, but start or finish that range on the extreme value of the range given in the display date?

• After XXXX-YYYY: YYYY(+1) -> YYYY(+5)
• After XXXXs: XXXX(+10) -> XXXX(+19), effectively the next decade.
• Before XXXX-YYYY: XXXX(-5) -> XXXX(-1)
• Before XXXXs: XXXX(-10) -> XXXX(-1), effectively the previous decade.
• After ABth century: AB00 -> AB50

Early, Mid and Late

• Early XXth century: XX00-XX25
• Mid-XXth century: XX25-XX75
• Late XXth century: XX75-XX99

• Early XXXXs: XXX0-XXX4
• Mid XXXXs: XXX3-XXX7
• Late XXXXs: XXX5-XXX9

• Early XXXX: XXXX-01-01 -> XXXX-4-30
• Mid XXXX: XXXX-05-01 -> XXXX-8-31
• Late XXXX: XXXX-09-01 -> XXXX-12-31

Others

• First Half, Second Quarter, Last third of the XXXXs or XXth Century: These are just translated into the implied year ranges. It might be easier for people using this data if these terms where simplified to Early, Mid and Late, but this is not really required and the terms are quite straight forward as is.
• There are a few examples in some non-NG objects of DisplayDate with Month and Day information, these are just converted into the standard syntax, YYYY-MM-DD.
• apparently dated XXXX: This one does not seem to be very helpful. This only for NG673 and there is an on going discussion about the meaning of the words in the inscription, quite ironic actually. It would probably be clear with probably XXXX, or even probably About XXXX or as sugested in the CIC text going to the decade with probably XXX0s.