SLIDE 1
35/123
A Simple Example
Pustejovsky - Brandeis Computational Event Models
36/123
The Final SDRS
Pustejovsky - Brandeis Computational Event Models
A Simple Example 35/123 Pustejovsky - Brandeis Computational Event - - PowerPoint PPT Presentation
A Simple Example 35/123 Pustejovsky - Brandeis Computational Event - - PowerPoint PPT Presentation
A Simple Example 35/123 Pustejovsky - Brandeis Computational Event Models The Final SDRS 36/123 Pustejovsky - Brandeis Computational Event Models Some Lexical Semantics 37/123 Pustejovsky - Brandeis Computational Event Models An Example
SLIDE 2
SLIDE 3
37/123
Some Lexical Semantics
Pustejovsky - Brandeis Computational Event Models
SLIDE 4
38/123
An Example of Narrative
Pustejovsky - Brandeis Computational Event Models
SLIDE 5
39/123
Minimal SDRS
Pustejovsky - Brandeis Computational Event Models
SLIDE 6
40/123
TimeML: Temporal Ordering of Events
Verhagen (2005), Pustejovsky (2017)
how a temporal closure component can be embedded in a temporal annotation environment. Temporal closure takes known temporal relations in a text and derives new implied relations from them, in effect making explicit what was implicit. A temporal closure component helps to create an annotation that is complete and consistent.
Pustejovsky - Brandeis Computational Event Models
SLIDE 7
41/123
Temporal Ordering of Events
Pustejovsky - Brandeis Computational Event Models
SLIDE 8
42/123
Temporal Ordering of Events
Pustejovsky - Brandeis Computational Event Models
SLIDE 9
43/123
Temporal Ordering of Events
Pustejovsky - Brandeis Computational Event Models
SLIDE 10
44/123
Temporal Ordering of Events
Pustejovsky - Brandeis Computational Event Models
SLIDE 11
45/123
Temporal Ordering of Events
Pustejovsky - Brandeis Computational Event Models
SLIDE 12
46/123
Temporal Ordering of Events
Pustejovsky - Brandeis Computational Event Models
SLIDE 13
47/123
Temporal Ordering of Events
Pustejovsky - Brandeis Computational Event Models
SLIDE 14
48/123
Computing Event Ordering
Events in isolation with their subevent structure (individual dynamic event graphs) Ordering over multiple events in text or discourse
Pustejovsky - Brandeis Computational Event Models
SLIDE 15
49/123
Same Information as a Labeled Transition System Graph
Pustejovsky - Brandeis Computational Event Models
SLIDE 16
50/123
Making Relation Annotation More Informative
Narrative Container is the default interval containing the events being discussed in the text, when no explicit temporal anchor is given. (1) Put events in temporal containers. (2) Order events relative to temporal anchors. (3) Some temporal containers may be implicit. (4) Temporal containers may be style or genre specific.
Pustejovsky - Brandeis Computational Event Models
SLIDE 17
51/123
Document Creation Time 1/2
10-26-1989 1 Philip Morris Co., New York, adopted a defense measure designed to make a hostile takeover prohibitively expensive. 2 The giant foods, tobacco and brewing com- pany said it will issue common-share purchase rights to shareholders of record Nov. 8.
Pustejovsky - Brandeis Computational Event Models
SLIDE 18
52/123
Document Creation Time 1/2
4-10-2011 Local officials reported yesterday that a car exploded in down- town Basra.
Pustejovsky - Brandeis Computational Event Models
SLIDE 19
53/123
TimeBank Annotation Style
DCT= t1, val=10-04-2011 t2 = yesterday, val=09-04-2011 e1 = report e2 = explode TLINK1 = before(e1,t1) TLINK2 = before(e2,t1) TLINK3 = includes(t2,e1)
Pustejovsky - Brandeis Computational Event Models
SLIDE 20
54/123
The Missing Temporal Relation
TLINK4 = includes(t2,e2) e2 = explode t2 = yesterday, val=09-04-2011
Pustejovsky - Brandeis Computational Event Models
SLIDE 21
55/123
Document Creation Time 2/2
9-02-1989 An Orlon spokesman said that the Board rejected Margo’s lat- est takeover bid.
Pustejovsky - Brandeis Computational Event Models
SLIDE 22
56/123
TimeBank Annotation Style
DCT= t1, val=09-02-1989 e1 = say e2 = reject TLINK1 = before(e1,t1) TLINK2 = before(e2,t1) TLINK3 = before(e2,e1)
Pustejovsky - Brandeis Computational Event Models
SLIDE 23
57/123
Missing Temporal Relations
Reference to a Default Narrative Container (DNC) t2 = DNC, val=09-02-1989 TLINK4 = includes(t2,e1) TLINK5 = includes(t2,e2) e1 = say e2 = reject
Pustejovsky - Brandeis Computational Event Models
SLIDE 24
58/123
Narrative Container
Narrative Container is the default interval containing the events being discussed in the text, when no explicit temporal anchor is given. Narrative Time is the current temporal anchor for events in a document, and can change as the reader moves through the narrative. Narrative Scope describes the timespan described in the document, with the left marker defined by the earliest event mentioned, and the right by the event furthest in the future.
Pustejovsky - Brandeis Computational Event Models
SLIDE 25
59/123
Narrative Container 2/3
April 25, 2010 7:04 p.m. EDT -t0 S1: President Obama paid-e1 tribute Sunday -t1 to 29 work- ers killed-e2 in an explosion -e3 at a West Virginia coal mine earlier this month- t2, saying-e4 they died-e5 “in pursuit of the American dream.” S2: The blast-e6 at the Upper Big Branch Mine was the worst U.S. mine disaster in nearly 40 years.
Pustejovsky - Brandeis Computational Event Models
SLIDE 26
60/123
Narrative Container 3/3
t2 "earlier this month" t1 "Sunday" e3 explosion e5 "died" e1 "paid" e2 "killed" e4 "saying" t0 DCT e6 "blast" t0 DCT t1 "Sunday" e2 "killed" t2 earlier this month e5 "died" e6 "blast" e1 "paid" e4 "saying" e3 explosion
A B
Figure: A: Times and events as appearing in the text; B: events grouped into their appropriate Narrative Times.
Pustejovsky - Brandeis Computational Event Models