Database Reference
In-Depth Information
Table 5. TimeML expression for “more likely 1 January 2000 than 2 January 2000”
<TIMEX3 tid=”t1” type=”DATE” value=”2000-01-01”>
On January 1st, 2000
</TIMEX3>
<CONFIDENCE tagType=”TIMEX3” tagID=”t1” confidenceValue=”1.”/>
<TIMEX3 tid=”t2” type=”DATE” value=”2000-01-02”>
On January 2nd, 2000
</TIMEX3>
<CONFIDENCE tagType=”TIMEX3” tagID=”t2” confidenceValue=”0.80”/>
The metadata provider should define the time indications of events and observations by means of a
metadata editor.
A running example of the metadata of two thematic maps representing subsequent observations of
the same event, i.e., the melting of the “
Lys Glacier
” (a glacier of the Italian Alps), during Summer
2007, is discussed.
In this example, the temporal specification of the first occurrence of the observation is known pre-
cisely, while the second date of the observation is affected by some uncertainty. Following the proposed
extension, in the metadata of the first map we have fields such as:
Metadata 1
…
Event=”Lys Glacier melting”
Occurrence=”observation of Lys Glacier melting”
Time Position of Occurrence=”1.7.2007”
…
In the metadata of the second map we have:
Metadata 2
…
Event=”Lys Glacier melting”
Occurrence=”observation of Lys Glacier melting”
Time Position of Occurrence=”3.9.2007” with confidence 0.7 or =”4.9.2007” with
confidence 0.3
…
The Metadata items are translated into TimeML sentences as in Table 6.
To allow partial matching with respect to soft selection conditions specified by a user, a parser trans-
lates the external TimeML definitions of temporal metadata element into its internal fuzzy set represen-
tation.
In this phase we can obtain fuzzy sets defined on distinct domains (G) having different time granularity.
On the other side, the user specifies her/his temporal selection conditions Q within a discovery service
client interface. The expression of temporal selection conditions could be performed by the support of
