matching: initial release

inspirehep / beard

Bibliographic Entity Automatic Recognition and Disambiguation

Other

66 stars 36 forks source link

Closed MSusik closed 8 years ago

MSusik commented 9 years ago

Introduces a matching algorithm for the results of disambiguation. The results are matched to the disambiguation that was available previously, before the run of BEARD.

Signed-off-by: Mateusz Susik mateusz.susik@cern.ch

glouppe commented 9 years ago

For further reference (no need to take this into account in this PR), here is another approach I did using a linear programming formulation instead of the hungarian algorithm to solve the matching problem. The possibility to add random constraints yields more freedom for modeling the problem. https://github.com/glouppe/notebooks/blob/master/Aligning%20clusters%20with%20linear%20programming.ipynb

MSusik commented 9 years ago

Your method probably should be added to the code as a replacement for munkres library

kaplun commented 8 years ago

@glouppe what is the status of this branch?

glouppe commented 8 years ago

Given the direction that @jacenkow has taken with invenio-beard, I think this branch will not be continued.