Loading the required date takes far more than a minute, does it have be so long? I wonder whether the term-pair-similarity.csv.gz file is extracted each time?
Remove all HP:0*s from term-pair-similarity.csv would also save much space - and the use of integers instead of strings might be faster anyway. The ic_mica could also become an integer (e.g. by multiplying it with 10,000 and round it).
No idea how Java deals with floats and strings, but in most programming languages and databases, this would save much RAM and CPU time.
Loading the required date takes far more than a minute, does it have be so long? I wonder whether the term-pair-similarity.csv.gz file is extracted each time?
Remove all HP:0*s from term-pair-similarity.csv would also save much space - and the use of integers instead of strings might be faster anyway. The ic_mica could also become an integer (e.g. by multiplying it with 10,000 and round it).
No idea how Java deals with floats and strings, but in most programming languages and databases, this would save much RAM and CPU time.