HapPenIng

Input Data (Event Graphs and Triples)

Title	Description	Paper Reference	Example	Size & Download
Wikidata Event Graph	Event Graph extracted from Wikidata containing 350K events, 9K event series, sub-event relations, follow-up event relations, event locations and event dates.	Section 4.1		63.8MB
Wikidata Event Graph (with new sub-event relations)	Wikidata Event Graph, enriched with newly generated sub-event relations.	Section 4.1		64.5MB
Wikidata Triples	All triples consisting of a subject entity, a property, and an object entity contained in Wikidata. They are used in HapPenIng to train an embedding model.	Section 4.1		6.7GB

Results (New Sub-Event Relations and Events)

Title	Description	Paper Reference	Example	Size & Download
New events	New events generated during the event inference procedure, using the ECW constraint.	Section 4.3		747.2KB
New sub-event relations	New sub-event relations inferred during the sub-event prediction procedure (using all features and the random forest classifier).	Section 4.2		7.1MB
Sub-event relations from event inference	New sub-event relations inferred during the event inference procedure, using the ECW constraint.	Section 4.3		610.2KB

We also provide results for the other HapPenIng configurations (ZIP file, 4.9MB compressed).

Evaluation Data (Cross-validation Folds and Corrupted Event Series)

Title	Description	Paper Reference	Example	Size & Download
Cross-validation folds for sub-event relation prediction	10 folds of training, test and validation sets with sub-event relations.	Section 3.1		10.6MB (compressed)
Removed events for reconstruction (corruption factor 5%)	A list of events that were removed from the Wikidata Event Graph to evaluate reconstruction coverage (5% of event series leaf nodes are removed).	Section 4.3		130.9KB
Removed events for reconstruction (corruption factor 10%)	A list of events that were removed from the Wikidata Event Graph to evaluate reconstruction coverage (10% of event series leaf nodes are removed).	Section 4.3		265.0KB
Removed events for reconstruction (corruption factor 15%)	A list of events that were removed from the Wikidata Event Graph to evaluate reconstruction coverage (15% of event series leaf nodes are removed).	Section 4.3		396.2KB

Annotated Samples (Sub-event Relations and Events)

Title	Description	Paper Reference	Example	Size & Download
Annotated samples of new sub-event relations	Random sample of 100 newly generated sub-event relations (learnt with all features and with STransE embeddings only) with manual annotations.	Section 4.2		18.8KB
Annotated samples of new events	Random sample of 100 newly generated events for each HapPenIng configuration (BSL, EVO, INT, WIN, COV, CWI, ECW) with manual annotations.	Section 4.3		73.0KB

Subject	Predicate	Object
Q31	P78	Q39773
Q31	P85	Q161539
Q31	P163	Q12990
Q31	P150	Q9337
Q31	P150	Q240
Q31	P150	Q9331

...

event	new_sub_event	wikidata_id	start_date	end_date	locations	date_source
2017 Peru International Series	2017 Peru International Series – Women's singles	http://www.wikidata.org/entity/Q30225700	2017-01-01	2017-12-32	null	3
2012 World Fencing Championships	Women's team foil at the 2012 World Fencing Championships	http://www.wikidata.org/entity/Q580944	2012-01-01	2012-12-31	null	2
2018 Rio Open	2018 Rio Open – Men's Singles	http://www.wikidata.org/entity/Q47467357	2018-01-01	2018-12-32	null	3

...

event: label of the new event's parent event
new_sub_event: label of the new event
wikidata_id ID: Wikidata ID of the new event's parent event
start_date: start date of the new event
end_date: end date of the new event
locations: locations of the new event
date_source: the rule that was used to infer the event time
(1: parent event time, 2: parent event year, 3: extracted from the label)

parent_event	sub_event	parent_wikidata_id	sub_event_wikidata_id
1988 Polish Badminton Championships	1988 Polish Team Badminton Championships	Q178530	Q16891513
2012 Carolina Challenge Cup	2012 Scottish Challenge Cup Final	Q4625092	Q4626709
2012 Carolina Challenge Cup	2012 IIHF Challenge Cup of Asia	Q4625092	Q4625815

...

parent_event: parent event
sub_event: sub event
parent_wikidata_id: Wikidata ID of the parent event
sub_event_wikidata_id: Wikidata ID of the sub event

Parent Event	Sub Event	Class
Q544995	Q25927277	0
Q613506	Q847280	0
Q9636	Q667101	1
Q13384406	Q56070731	0
Q26838733	Q27627755	1
Q3534831	Q3965987	1

...

Class: "1", if correct sub-event relation. 0 otherwise.

Wikidata ID	Label
Q3992328	1982 Tokyo Indoor – Doubles
Q27880114	1970 Mexican Badminton Championships – Mixed doubles
Q191152	Zentropacup
Q27669316	1992 Welsh International Badminton Championships – Women's doubles
Q47605556	2016 World Rowing Junior Championships – Junior men's coxless four

...

Removed events for reconstruction, with Wikidata ID and label.

No.	Rating	Parent event	Sub event	Parent Event Wikidata ID	Sub Event Wikidata ID
1	1	National Professional Soccer League	1990–91 National Professional Soccer League season	Q6975029	Q4586572
2	0	1997 NHK Trophy	1997 SAS Badminton Trophy	Q2582720	Q1307530
3	1	2010 Asian Men's Cup Volleyball Championship	2010 Asian Junior Men's Volleyball Championship	Q2421442	Q4616638

...

Rating: manual annotation ("1", if there is a sub-event relation, "0" if not)
Parent Event: label of the parent event
Sub event: label of the sub event
Parent Event Wikidata ID: Wikidata ID of the parent event
Sub Event Wikidata ID: Wikidata ID of the sub event

No.	Rating	Parent event	New sub event	Wikidata ID	Start date	End date	Locations	Date source
1	0	1964 Summer Olympics	Equestrian at the 1964 Summer Olympics – Mail coach	http://www.wikidata.org/entity/Q8420	1964-01-01	1964-12-31	null	2
2	1	2020 Summer Paralympics	Netherlands at the 2020 Summer Paralympics	http://www.wikidata.org/entity/Q14856650	2020-01-01	2020-12-31	null	2
3	1	2009 Ordina Open	2009 Ordina Open - Men's	http://www.wikidata.org/entity/Q1506655	2009-01-01	2009-12-31	null	2
4	1	United States presidential election, 1892	United States presidential election in Illinois, 1892	http://www.wikidata.org/entity/Q72472	1892-01-01	1892-12-31	null	2

...

Rating: manual annotation ("1", if the new event represents a real-world event, "0" if not)
Parent Event: label of the new event's parent event
New sub event: label of the newly inferred event
Wikidata ID: Wikidata ID of the new event's parent event
Date source: the rule that was used to infer the event time
(1: parent event time, 2: parent event year, 3: extracted from the label)