ItsdbTreebanking_ItsdbExporting

Exporting Trees

You can output treebanked data in other formats with Trees | Export. This outputs the data as one gzipped file per item.

If you set the switch to thinning normalize, the system only outputs results for selected (active) trees (Trees | Export | Thinning Export or (setf tsdb::*redwoods-thinning-export-p* t)).

Possible output formats are listed below:

• derivation---derivation tree: primary, labeled in terms of grammar-internal identifiers;

• tree---phrase structure tree: derived, labeled using a set of abbreviatory symbols;

• avm---attibute value matrix: derived, the full HPSG sign, including all daughters;

• mrs---MRS: meaning representation, raw;

• indexed---MRS: indexed;

• prolog---MRS: prolog style;

• rmrs---Robust MRS: meaning representation as RMRS;

• xml---Robust MRS: formatted with XML;

• dependencies---dependencies: derived, elementary dependency relations (reduced form of MRS);

  • print all relations with (setf mrs::*eds-include-vacuous-relations-p* t)

• triples ---dependency triples: PRED ARG1 PRED

• all---All the representations.

You can set what information gets output in '.tsdbrc' (ItsdbCustomization).

(setf tsdb::*redwoods-export-values* '(:derivation :tree :mrs :prolog))

Note: Some of these data structures may require reconstruction. In this case you need to have the same version of the grammar loaded (in the LKB) that was used to parse the profile.

Note: To get the cfrom/cto working in xml and dependencies you need to set lkb::*characterize-p* to non-null before you export.

(setf lkb::*characterize-p* t)

Exporting can be memory intensive. Use a batch instead for large profiles. There is an example described in RedwoodsTop.