> Is there a better heuristic for deciding whether to use Commentz-Walter?

No, I have no idea.

> Can the Galil rule be adapted to Commentz-Walter?

No, I don't know even whether we can do it.

> Another possibility might be to replace Commentz-Walter with some other 
> algorithm (Aho-Corasick, modified Wu-Manber, etc.).

DFA is already like Aho-Corasick.  So even if we imprement them, there
will be no merit much in spite of big changes.

The argument has convinced me that we shouldn't apply the patch now.

Thanks,
Norihiro




Reply via email to