summaryrefslogtreecommitdiff
path: root/src
AgeCommit message (Collapse)AuthorFilesLines
2014-06-11getBrands: increase worker count from 2 to 4Peter Wu1-1/+1
Performance details: - 1 worker : 62k/min (n=60, sd=5.6k, max=70k) - 2 workers: 97k/min (n=45, sd=12k, max=123k) (+35k) - 3 workers: 133k/min (n=39, sd=23k, max=161k) (+36k) - 4 workers: 190k/min (n=102,sd=38k, max=230k) (+57k) No idea where the +57k spike comes from.
2014-06-11Use more workers for consumersPeter Wu1-8/+33
This improves the performance from 60k to about 95k tweets per second.
2014-06-11Limit updating to unclassified tweetsPeter Wu1-1/+4
Tweets which do not mention a brand at all will have a brand "no".
2014-06-11Commit results early to avoid data loss on abortPeter Wu3-7/+26
2014-06-11More numbers for estimating durationPeter Wu2-0/+23
2014-06-11getBrands: add periodic status reporting, save memoryPeter Wu4-31/+158
Prevent out of memory error due to buffering the full resultset. This is done by disabling auto commit (starting a transaction) and setting a smaller fetch size. Watcher is based on my work in Datafiller.
2014-06-11Change to use 'RESET' instead of booleanPeter Wu2-23/+11
Also make selection query smaller
2014-06-04Producer / consumer for getBrandsPeter Wu2-17/+102
2014-06-04Formatting, remove unused importsPeter Wu2-123/+116
2014-06-04timezone command occurred twice.Maurice Laveaux1-1/+0
2014-06-04Merge remote-tracking branch 'origin/master'Maurice Laveaux2-8/+183
Conflicts: src/main/Analyzor.java src/main/FarmShell.java
2014-06-04Added better notifications on getBrands.Maurice Laveaux2-585/+643
* Reports amount of tweets processed.
2014-06-02improved ratinginterval in posnegs1231881-1/+3
2014-06-02+5s1231881-1/+1
2014-05-31made a method that makes a csv to show the newsspreading process in disco. ↵s1231882-2/+110
Only looks at the replies/retweets of one tweet for a case, no tree by recursion (don't know how to show this in disco). Still needs testing.
2014-05-31made a method for the posnegVisualizers1231881-7/+71
2014-05-28Merge origin/masterdaanpeters5-607/+609
Conflicts: src/main/Analyzor.java src/main/FarmShell.java
2014-05-28Changed Timezonedaanpeters2-346/+372
2014-05-27changed Analyzor.timezone(String query) so that it adds a legenda "timezone" ↵s1231881-349/+361
so that the visualizer can display a legends
2014-05-26Formatting, drop useless license headerPeter Wu3-57/+40
Sam Hocevar doesn't know us nor does he have copyright about this work...
2014-05-26Optimize sentiment analysisPeter Wu3-16/+39
Do not create a new prepared statement every time. Use an optimized UPDATE query. Drop requirement for supplying a query.
2014-05-26FKING CRLFPeter Wu1-345/+345
2014-05-21Made a method in Analyzor that makes a csv file for an upcoming html app ↵s1231881-296/+345
that shows the amount of users in a timezone about a brand. Also made printing a csv file from a String, String, Integer hashMap a seperate method.
2014-05-21Change wordcloud with added brandsS1297781-285/+296
2014-05-19fixed wordcloud agains1231881-3/+3
2014-05-19fixed wordclouds1231881-1/+2
2014-05-19better value format for wordcloud.csvs1231881-1/+1
2014-05-19better header for wordcloud.csvs1231881-1/+1
2014-05-19added header for wordcloud.csvs1231881-282/+284
2014-05-19Merge branch 'master' of git@git.lekensteyn.nl:tue/2IOC0-DBL/Goldfarmer.gitunknown1-14/+27
Conflicts: src/main/Analyzor.java
2014-05-19wordcloud now makes csv files for wordcloud.htmlunknown1-12/+14
2014-05-19double exception fixedS1297781-18/+31
2014-05-19BrandChecker removePunct replaces everything.Maurice Laveaux1-2/+2
2014-05-19Also remove @ and # from input text.Maurice Laveaux1-2/+1
2014-05-19Replace the punctuality with spaces.Maurice Laveaux1-1/+1
2014-05-19Added javadoc and visibility parameters.Maurice Laveaux1-8/+29
2014-05-19Merge origin/masterS1297781-23/+68
2014-05-19Merge origin/masterS1297781-245/+246
Conflicts: src/main/Analyzor.java
2014-05-19Merge branch 'master' of git.lekensteyn.nl:tue/2IOC0-DBL/GoldfarmerMaurice Laveaux3-264/+255
2014-05-19Updated the brandchecker code and rules.Maurice Laveaux1-23/+68
* Changed the brandrules.txt ruleset to conform the new parsing of rules. * Brandchecker now doesn't count multiple equal words. * Updated the testcases, which don't work for others.
2014-05-19Merge origin/masterS1297781-245/+245
Conflicts: src/main/Analyzor.java
2014-05-19get brands connection to databaseS1297783-228/+255
2014-05-19Added some javadoc and visibility to Analyzor.Maurice Laveaux1-228/+264
2014-05-19Merge origin/masterS1297781-221/+228
Conflicts: src/main/Analyzor.java
2014-05-19changed disco outputS1297781-1/+1
2014-05-18wordcloud now outputs a csv file with "tweetid, word" to be used in disco. ↵s1231881-23/+30
When the brandchecker works, wordcloud can use it to divide it into brands.
2014-05-15changed disco outputS1297781-0/+4
2014-05-15added disco outputS1297782-1/+29
2014-05-15Merge branch 'master' of git.lekensteyn.nl:tue/2IOC0-DBL/GoldfarmerMaurice Laveaux2-216/+201
Conflicts: src/main/Analyzor.java
2014-05-15Added first version of BrandCheckerMaurice Laveaux2-208/+334
TODO * Not yet finalized, need to search through ,.;'/[] etc. * Maybe implement searching after eachother, first htc then one.
generated by cgit v1.2.1 (git 2.18.0) at 2024-06-13 02:02:23 +0000