##################
#Initial Seed Set#
##################

We run our approach on Wikipedia articles with the same seed set that is used to extract ownership_of relationship on New York Times corpus.
The initial seed set contains five random company pairs as follows.
1.	AOL	Netscape
2.	Viacom	Viacom Media Networks
3.	Ford Motor Company	Jaguar Cars
4.	Time Warner	Turner Broadcasting System
5.	General Electric	NBC Sports

##################
#Approach Setting#
##################

We applied our algorithm with different parametrizations, which are the same as the New York Times corpus.
In this example result, we publish one of them that is using Hit score and select top 10 patterns and seeds.

###################
#Selected Patterns#
###################
	key-phrases		direction
1.	unit of			<-
2. 	acquisition of	<-
3.	subsidiary of	<-
4.	division of		<-
5.	owned by		<-
6.	owned			->
7.	acquired by		<-
8.	acquired		->
9.	company of		->
10.	merger with		<-
11.	sold			<-

################
#Selected Seeds#
################

1.	Viacom	Nick Records
2.	Ford Motor Company	Jaguar Cars
3.	General Electric	NBC Sports
4.	Comcast	NBC Sports
5.	Oracle Corporation	Sun Microsystems
6.	Viacom	Viacom Media Networks
7.	Time Warner	Warner Bros.
8.	Microsoft	Nokia
9.	CBS Corporation	CBS Television Stations
10.	Rogers Communications	Rogers Media
11.	Cox Enterprises	Cox Media Group
12.	AOL	Netscape
13.	Time Warner	Turner Broadcasting System

#########################
#Extracted Company Pairs#
#########################
In total 4722 distinct company pairs are extracted
Please find the detailed list in the file Wikipedia_OwnershipOf_ExtractedCompanyPairs.csv

##########
#Comments#
##########
Since Wikipedia articles are encyclopedic ones, the expectation of the same company pair is repetitively mentioned together is relatively low. 
Therefore, the frequencies of each company pairs that are extracted by our approach are also relatively low.    
