SentiStrength performance tests

Test 1:

Reading a large file of tweets, classifying the tweets and saving the results.

Commands and output

Started 23/10/2012 - 21:17:36.49
C:\Users\Mike>java -jar SentiStrength.jar sentidata f:/SentStrength_Data/ input
Ended 23/10/2012 - 21:24:37.91


Computer used for test

64 bit Windows PC, Intel i3 560 @ 3.33 Ghz, 16.0 GB RAM. 64 bit Windows PC, Intel i3 X5650 @ 2.67 Ghz, 96.0 GB RAM (2 processors).

Time to read file, classify sentiment and save results.

421 seconds

428 seconds

Texts processed

6,999,062 lines

6,999,062 lines

Processing speed 1

16,625 lines/sec

16,353 lines/sec

Mb in file

752 Mb

752 Mb

Processing speed 2

1.78 MB/s

1.76 MB/s



SentiStrength will be MUCH slower if it is invoked separately for each text fed into it.

The above includes reading the input file from disk and saving the results to disk. It should be a little quicker if the texts have already been read in and are fed to it.