Future Plans

1.

Finding a faster implementation of the min-cut algorithm (for example [10]).

2.

Improving the algorithm by attempting to find maximal highly connected subgraphs (see e.g. [13]) with a minimal complexity damage.

3.

Trying different similarity scoring schemes.

4.

Trying different objective functions.

5.

Finding ways for measuring the quality of a solution when the true clustering is unknown.

6.

Using cluster fingerprint to find database matches.

7.

Use the HCS clustering in other contexts.