I found a problem when using bulk similarity computation to get kgkt scores.
When the number in the input pair is larger than 20, the output scores will be more than 20.
In one case I countered, my input size is 200 while the output I receive is about 900.
For your reference, I have attached the TSV files of input and output.
Hello,
I found a problem when using bulk similarity computation to get kgkt scores.
When the number in the input pair is larger than 20, the output scores will be more than 20. In one case I countered, my input size is 200 while the output I receive is about 900.
For your reference, I have attached the TSV files of input and output.
Thank you.
Sincerely,
Zhiyong test_data.zip