License findings are not correct

[GMishx]

License findings are not correct

[amanjain97]

Approach: Excluded the top 100 common English words from the extracted source code comments to minimize the chance of matching with licenses with more common English words.
But this is not making much difference with the classification. Please suggest other approaches too.

[ag4ums]

Another approach sugession:
to identify the file is license relevant or not using a set of keywords( may be a list combining all the license text)
then, we can take the license relevant text from nomos (STRING.in) to find how much match a known file gives.
How it will be different from nomos : the license signature can be easily updated and new signature can be added dynamically.

[ag4ums]

or we can try with other probabilistic approaches.

[amanjain97]

We shall maintain this in wiki/drive documents that what algorithm should be used in order to get best results