This repo includes different MMCS algorithms implemented in Java for discovering minimal hitting sets.
Currently it includes:
-
MMCS: Original MMCS. Located in algorithm/MMCS
-
EMMCS: Extendable MMCS that supports inserting new subsets that should be covered. Located in algorithm/EMMCS
-
BHMMCS: Bi-directional Hybrid MMCS that supports both inserting and removing subsets. Removing is implemented by re-initiating from the remained subset, which is currently the fastest method. Located in algorithm/BHMMCS
Additionally, it includes corresponding FD Connectors for each MMCS algorithm (located in algorithm/fdConnectors), which provide convenience for discovering minimal FDs.
Mmcs mmcs = new Mmcs(nElements); // nElements is the number of total attributes of input dataset
mmcs.initiate(diffSets); // diffSets, in form List<BitSet>, represents subsets to be covered
List<BitSet> minCoverSets = mmcs.getMinCoverSets(); // retrieve the resulted minimal hitting sets
Emmcs emmcs = new Emmcs(nElements);
emmcs.initiate(diffSets);
List<BitSet> minCoverSets0 = mmcs.getMinCoverSets();
emmcs.insertSubsets(insertedSets1); // insertedSets, in form List<BitSet>,
emmcs.insertSubsets(insertedSets2); // represents newly inserted subsets to be covered
...
List<BitSet> minCoverSets1 = emmcs.getMinCoverSets(); // retrieve the resulted minimal hitting sets
Bhmmcs bhmmcs = new Bhmmcs(nElements);
bhmmcs.initiate(diffSets);
List<BitSet> minCoverSets0 = bhmmcs.getMinCoverSets();
bhmmcs.insertSubsets(insertedSets1);
bhmmcs.removeSubsets(remainedSets2); // remainedSets2 represents remained subsets after removal
...
List<BitSet> minCoverSets1 = emmcs.getMinCoverSets(); // retrieve the resulted minimal hitting sets
Basically the same as their corresponding MMCS algorithms. The only difference is that an FdConnector will successively call its MMCS algorithm on each RHS and return a list of minimal FDs on all RHSs.