Hi Silvano, We are always interested in making the license detection more useful and accurate. There are two deficiencies in nomos. One is that it doesn’t give a confidence score (like the license is a 100% match), the other is that it doesn’t show you where the differences are between the reference license and the license found. If you look in the license_file table you will see fields for recording that information that nomos leaves blank. For a while now I’ve thought that the best solution might be to include Daniel German’s Ninka scanner and use it to complement nomos. Your agent would be another possibility.
Yes, please continue this discussion and put your draft implementation on fossology-devel. Thanks, Bob Gobeille On Jan 13, 2014, at 7:34 AM, Cirujano Cuesta, Silvano <[email protected]> wrote: > Hi Larry, > > My intention with this message was just getting a feeling for the interest in > the contribution of such an improvement. > > The problem that we see is that the signatures don't cover the whole licenses > (in case of long licenses even less that 10% of the text). That's why we say, > that Nomos gives only "hints to licenses". If Nomos findings are accepted > without review, then an assumption is being taken: the whole license text is > there although only a part of it has been found. This assumption will be true > in most of the cases, but now always. Since we want to automate the posterior > processing of the findings, we require more security on the presence of a > reported license. > > We wouldn't like to misuse the current Nomos implementation exploding > signatures for full license matching because of performance reasons. We are > creating a new agent for this task. > > I planned to give a draft of the implementation we are already working on in > the "developers" list. Is it right for you? > > Regards, > Silvano > > -----Original Message----- > From: Shi, Yao-Bin (Larry, GBU-HPSW, Cloud and OS) > [mailto:[email protected]] > Sent: Montag, 13. Januar 2014 16:25 > To: Cirujano Cuesta, Silvano; [email protected] > Subject: RE: Improve findings accuracy > >> The solution that we have in mind is to use the license texts that are > [larry] for me, that means we should give a more concrete(accurate) signature > to match, right? > Certainly if the signature is one license's whole text, that is almost > impossible to match any licenses in the scanning files. > More discussion is needed. >> stored in licenseref.sql to check if the whole text can be found (what >> we call a 100%-match). > _______________________________________________ > fossology mailing list > [email protected] > http://lists.fossology.org/mailman/listinfo/fossology _______________________________________________ fossology mailing list [email protected] http://lists.fossology.org/mailman/listinfo/fossology
