The Coreference Solver module reads this file to find out some needed parameters. The file has three sections:
<ABModel>
path to the file containing the
trained AdaBoost model. The .abm file contains an AdaBoost
model based on shallow Decision Trees (see [CMP03] for
details). You don't need to understand this, unless you want to
enter into the code of the AdaBoost classifier.
The name of the file may be either absolute or relative to the
location of the Coreference Solver config file.
e.g:
<ABModel> coref.abm </ABModel>
It may be generated from an annotated corpus using the training programs that can be found in src/utilities/coref, which use the libomlet package. (see http://www.lsi.upc.edu/~ nlp/omlet+fries)
If you need to know more about this (e.g. to develop a Coreference Solver for your language) please contact FreeLing authors.
<SemDB>
specifies the files that contain a
semantic database. This is required to compute some WN-based attributes
used by the solver.
The section must contain two lines specifying two semantic information files, a SenseFile and a WNFile. The filenames may be absolute or relative to the location of the dependency rules file. For example:
<SEMDB> SenseFile ../senses16.db WNFile ../../common/wn16.db </SEMDB>
<MaxDistance>
states the maximum distance (in
words) at which possible corefernces will be considered. Short
values will cause the solver to miss distant coreferents.
Long distances will introduce a huge amount of possible coreferent
candidate pairs, slow the system, and produce a larger amount of
false positives.
Lluís Padró 2010-09-02