source build logs
interesting, most open source, majoriyty from travis or travistorrent, other sources not really present
number
look at big numbers: infeasible to do manual
kind of retrieved:
if you can do chunks you can do classification -> better to do chunks
info targeted:
mainly errors => that’s why we chose these logchunks for our data set
name of implementation:
no graph, only numbers in text
not much reuse, also not many published implementations
Technique:
combined with technique explained / detail
often very difficult/had to guess which technique is used. e.g. „expression-based“ string analysis
alles was 1/2x erwähnt ist fliegt raus
source build logs interesting, most open source, majoriyty from travis or travistorrent, other sources not really present
number look at big numbers: infeasible to do manual
kind of retrieved: if you can do chunks you can do classification -> better to do chunks
info targeted: mainly errors => that’s why we chose these logchunks for our data set
name of implementation: no graph, only numbers in text not much reuse, also not many published implementations
Technique: combined with technique explained / detail often very difficult/had to guess which technique is used. e.g. „expression-based“ string analysis alles was 1/2x erwähnt ist fliegt raus
Impl avail: one third was travistorrent