Each video was annotated 5 times. 
For the matching and ranking subtask, these were divided into 5 sets (A, B, C, D, E). 
The groundtruth file, Matching_Ranking_GT.txt, is organized as follows:

        1st column: video_URL_ID
        2nd column: Matched Textual description from set A
        3rd column: Matched Textual description from set B
	4th column: Matched Textual description from set C
	5th column: Matched Textual description from set D
	6th column: Matched Textual description from set E