Experimental Results

Fingerprint Indexing is a new, simple but practically efficient non-perfect term indexing technique. It supports retrieval of terms matching a query term, matched by the query, or unifiable with it, with uniform, simple algorithms.

Fingerprint Indexing can be seen as a variant of non-perfect discrimination tree indexing, or as a generalization of top-symbol hashing. It represents terms by fingerprints, constant-lengths vectors of samples of terms at (potential) positions. Fingerprints are organized in tries, and terms are stored at the leaves of the trie. Retrieval follows all compatible branches at each fork.

Fingerprint indexing is described in Fingerprint Indexing for Paramodulation and Rewriting.

1. •Short version published in the Proceedings of the 6th IJCAR
   

2. ‣Paper (306 kb PDF)
   

3. ‣Abstract
   

4. ‣BibTex
   

5. •Extended version
   

6. ‣Paper (340 kb PDF)
   

7. ‣Abstract
   

8. ‣BibTeX
   

9. •Final test runs were performed with development versions of E 1.4. All can be reproduced with E 1.4-19.
   

10. ‣E 1.4-19 source (~2.4 MB, .tgz)
    

11. •Test runs were done on the University of Miami Pegasus Cluster
    

12. ‣Jobs were run on 2.4 GHz 8 Core Intel Xeon machines with 16 GB of main memory each
    

13. ‣8 concurrent processes were scheduled per core
    

14. ‣A CPU time limit of 300 seconds and a memory limit of 512 MB were in force for each individual test run
    

15. •The tests were run on all CNF and FOF problems from TPTP-5.2.0.
    

16. •The results are stored in protocol files
    

17. ‣Each protocol contains results for all TPTP problems for a given parameter setting
    

18. ‣Lines starting with a # are comment lines
    

19. ‣In particular, the first line gives the command line options for the test run
    

20. ‣Lines starting with a problem name contain data for that particular problem
    

21. ‣The important columns are described in the legend.txt file.
    

22. •The results: EFPRes.tgz (~4.2 MB)