I'm a little surprised that 1 trillion comparisons is computationally infeasible.Why? 1GHz is one billion cycles per second, so 1 trillion CPU instructions will take about 10 minutes. Given that a string comparison with multiple function calls will take at least hundreds or thousands of CPU cycles, this will take at least a day or so. Probably a lot more.
If you want to do fast, approximate comparison of strings, a lot of work has been done on that in bioinformatics. In particular, the popular BLAST and PSI-BLAST algorithms more or less solve this problem by creating an index of short substrings, then doing a full comparison where they match. You might try to develop something similar.
In reply to Re^2: Cluster a big bunch of strings
by educated_foo
in thread Cluster a big bunch of strings
by citromatik
| For: | Use: | ||
| & | & | ||
| < | < | ||
| > | > | ||
| [ | [ | ||
| ] | ] |