CompareString2

T-UI 6.12 introduced CompareString2, a Java library which allows to perform String comparison with various algorithms.

Algorithms

In order to change the algorithm used to compare your Stings, you need to set the option suggestions_algorithm in suggestions.xml.

Choose an ID from the table below:

Category	Algorithm	ID
Distance	LCS	0
Distance	OSA	1
Distance	QGRAM	2
Normalized distance	COSINE	4
Normalized distance	JACCARD	5
Normalized distance	JAROWRINKLER	6
Normalized distance	METRICLCS	7
Normalized distance	NGRAM	8
Normalized distance	NLEVENSHTEIN	9
Normalized distance	SORENSENDICE	10
Normalized similarity	COSINE	11
Normalized similarity	JACCARD	12
Normalized similarity	JAROWRINKLER	13
Normalized similarity	NLEVENSHTEIN	14
Normalized similarity	SORENSENDICE	15
Metric distance	DAMERAU	16
Metric distance	JACCARD	17
Metric distance	LEVENSHTEIN	18
Metric distance	METRICLCS	19

Then use the following command:

config -set suggestions_algorithm ID

For instance, if you want to use the normalized-distance version of JACCARD, you will use the command:

config -set suggestions_algorithm 5

You can get more info about the available algorithms here and here.

Deadline

Result ranges

Category	Equals	Different
Distance	0	+Infinity
Normalized distance	0	1
Normalized similarity	1	0
Metric distance	0	+Infinity

As you can see, there are some cases when an higher result means that the Strings are "more equal" (i.e. normalized-similarity) whereas the situation is opposite in other cases (i.e. distance, normalized-distance, metric-distance).

Keep the table above in mind when you set your deadline, with the command

config -set suggestions_deadline [deadline]

Check here and here for more details.