TIMEX at MUC-6

The Sixth Message Understanding Conference (MUC-6) introduced the recognition of time expressions as part of the larger goal of named entity recognition (NER) in text. Two kinds of time expressions were required to be identified in MUC-6: dates and times. Since this was the first evaluative exercise of its kind, the goals set were relatively simple: the scope was restricted to absolute temporal references only; the internal structure of complex time expressions did not need to be identified; and there was no requirement to interpret (normalise) the expressions to any universal time.

Annotations of temporal expressions were inline and marked up with a TIMEX SGML tag, which had one attribute named type with possible values of date and time. Details of the guidelines can be found at www.cs.nyu.edu/cs/faculty/grishman/NEtask20.book_11.html. The main points were:

1. Only absolute time expressions referring to a specific segment of time were to be tagged.
2. Absolute time expressions were to be tagged as single items; subparts were not to be tagged separately, even in the case of possessive constructions where both subparts refer to some time.
   Examples:
   <TIMEX TYPE="TIME">twelve o’clock noon</TIMEX>
   the <TIMEX TYPE="DATE">autumn</TIMEX> report
   <TIMEX TYPE="DATE">fiscal 1989’s fourth quarter</TIMEX>
3. Relative time expressions were not to be tagged, but any absolute times expressed as part of the entire expression were to be tagged.
   Examples:
   <TIMEX TYPE="DATE">July</TIMEX> last year
   next <TIMEX TYPE="DATE">summer</TIMEX>
   thirty days before the end of the year [no markup in this case]
4. Special days, such as holidays, that are referenced by name, were not required to be tagged. They would be marked optional in the answer key; for example: because of the observance of <TIMEX TYPE="DATE" STATUS="OPT">All Saints’ Day</TIMEX>
5. In rare cases, multiword strings that were to be tagged as TIMEX would contain LOCATION (ENAMEX tag) substrings that were also to be tagged; for example: <TIMEX TYPE="TIME">1:30 p.m. <ENAMEX TYPE="LOCATION">Chicago</ENAMEX> time</TIMEX>

The following table gives an overview of the types of time expression that were considered in and out of scope for MUC-6.

Rule	In-scope	Out-scope
An expression of minutes must indicate a particular minute and hour.	20 minutes after 10	a few minutes after the hour 20 minutes after the hour
An expression of hours must indicate a particular hour.	midnight twelve o’clock noon noon	mid-day morning
An expression of days must indicate a particular day.	Monday 10th of October	first day of the month
An expression of seasons must indicate a particular season.	autumn	next season
An expression of financial quarters or halves of the year must indicate which quarter or half.	fourth quarter first half
An expression of years must indicate a particular year.	1995	the current year
An expression of decades must indicate a particular decade.	1980s	the last 10 years
An expression of centuries must indicate a particular century.	the 20th century	this century