Documentation Center

Example of how capitalization differences affect match score

The following table explains how the match score is computed when the lookup and hit segments differ only in that some words are capitalized in one and not the other. Each table represents a different capitalization penalty.

tm_score_capitalization_penalty=0.01 (the default)
Lookup SegmentHit SegmentDifferenceScoreScore derivation explanation
"This is a segment""This is a Segment"Hit segment has one word capitalized that Lookup didn???t.99.75%The one capitalized word difference is penalized 1%, and that word is 25% (one of four words) of the score. So, 3 of 4 match (totaling 75%) and the other 25% is penalized 1% (so that word would be 25 - .25 = 24.75%). So the total score would be 75% + 24.75% = 99.75%.
"this is a segment""This Is A Segment"Hit segment has four capitalized words, the Lookup segment none..99%This penalty will be for each of the 4 words, so the score will be 4 x 24.75 = 99%.
tm_score_capitalization_penalty=0.2
Lookup SegmentHit SegmentDifferenceScoreScore derivation explanation
"This is a segment""This is a Segment"Hit segment has one word capitalized that Lookup didn???t.95%The one capitalized word difference is penalized 20%, and that word is 25% (one of four words) of the score. 20% x 25% = 5%. So the score is 100% - 5% = 95%.
"this is a segment""This Is A Segment"Hit segment has four capitalized words, the Lookup segment none..80%This penalty will be 4 times the previous one, so it would be 20%; so the score would be 100% ??? 20% = 80%.