Investigate In-Close

[garfieldnate]

FCA has a lot more research and work put into it than AM. Perhaps the state-of-the-art algo, In-close 4 by
Simon Andrews, could be applied to speed up AM. I would try a three-phase implementation:

1. Run In-close 4 to get the concept lattice
2. Mark heterogeneous contexts in the lattice
3. Assign counts to the homogeneous contexts

[garfieldnate]

Simons has published an algorithm for In-Close5, now: https://shura.shu.ac.uk/21248/1/Andrews-NewMethodForInhereting%28AM%29.pdf. For whatever reason, only In-Close4 is available from the linked SourceForge page, though: https://sourceforge.net/projects/inclose/files/