felixneutatz / githubrepositoryclassifier Goto Github PK

Hi @kevinkepp and @FelixNeutatz. From your README.md it is not really clear which of the steps are absolutely required to classify our golden data set. Are the extractor / attachment download steps optional or mandatory? We do not like to do any training but just run a single command where to put in the file as input. Providing a docker image / virtual appliance or hosted web site where your program is already running would be optimal. The other solutions evaluated so far could be setup within minutes and the longest one took about 60 minutes to complete the classification of the 221 repositories we test with (with one token). Hearing that the Linux kernel (which is among the samples) already takes 12 hours may not make the evaluation of your solution suitable for most jury members. On the other hand we really love your documentation and would be thrilled to see your accuracy / precision on our data. If there was any way for you to provide us a setup where we do not have to specify 20 parameters, run many programs and have to wait multiple days before we see a result, you are still in the game (provided that your solution has great accuracy / precision)👍

Best, Johannes