Interactive Labeller - Does not match at probability 0.5

[sonalgoyal]

During model building of 5M NC dataset, saw some messages on the interactive learner - what does do not match with probability 0.5 mean?

[navinrathore]

The typical examples of the concerned statement are (with complete score value)

Zingg predicts the above records DO NOT MATCH with a similarity score of 0.000000
Zingg predicts the above records DO NOT MATCH with a similarity score of 0.361295
Zingg predicts the above records DO NOT MATCH with a similarity score of 0.486515
Zingg predicts the above records DO NOT MATCH with a similarity score of 0.499993
Zingg predicts the above records MATCH with a similarity score of 1.000000
Zingg predicts the above records MATCH with a similarity score of 0.518476
Zingg predicts the above records MATCH with a similarity score of 0.999999

As, currently in code, the precision for the "similarity score" is set to two decimal points. the value is rounded. e.g, 0.499993 gets converted to 0.50. Hence, on console, the message says "... above records DO NOT MATCH with a similarity score of 0.50".

[sonalgoyal]

please make it a % score to 2 dec places then

[sonalgoyal]

or rather do this https://stackoverflow.com/questions/1305827/prevent-round-off-in-string-format-2f-doublevalue-in-java?