Important use-cases and command line arguments / web service switches are presented below. The list is not complete! See fminer README and web services README for a more complete overview.
BBRC
| Use-Case | Command Line |
Web Service |
API |
| n-Frequent Subgraph Mining (w/o adding activities) | -f n | - | SetMinfreq(n) SetChisqActive(false) |
| n-Frequent Subgraph Mining (w added activities) | -d -b -f n -p 0.0 |
- | SetDynamicUpperBound(false) SetBackbone(false) SetMinfreq(n) SetChisqSig(0.0) |
| n-Frequent, Class-Correlated (p-value as lower bound) Subgraph Mining |
-d -b -f n -p p |
backbone=false min_frequency=n min_chisq_significance=p |
SetDynamicUpperBound(false) SetBackbone(false) SetMinfreq(n) SetChisqSig(p) |
| Backbone Refinement Class Mining (p-value as lower bound) | -f n -p p |
min_frequency=n min_chisq_significance=p |
SetMinfreq(n) SetChisqSig(p) |
LAST-PM
| Use-Case | Command Line |
Web Service |
API |
| Latent Structure Pattern Mining | -f n | min_frequency=n | SetMinfreq(n) |
Note:
- In BBRC, two modes for n-frequent subgraph mining are available. The first one does not use activity information, whereas the second one uses it to indicate occurrences of subgraphs per class (for nominal activities), which can be interesting despite deliberately not taking class correlation into account during the mining process.
- For all settings (except web service) environment variables play a major role for the output format. Supported are gSpan and SMARTS, the latter as CSV or YAML. Moreover, p-values can be requested instead of test statistics. A special case is the output of hit counts. While on the command line, there is a special environment variable, the webservice must be passed a special switch: -d “nr_hits”.
