|
Class Summary |
| ConstantWS |
Using a constant weight makes the weighting word order independent. |
| MangesWS |
Calculates the weighting based upon the distance to the current label in
the following manner: weight=(2^(1-distance to focus word)). |
| MartinsWS |
Calculates the weighting based upon the distance to the current label in
the following manner: weight=(1/distance to focus word)
but also gives a higher weight to content words. |
| PermutatingWS |
Instead of weighting the context labels e.g. |
| RosellsWS |
((1-c) + c*sim) * 2^(1-d)
Note: this will not produce reproducible results in a multi-threaded environment
due to the TermFrequency constraint (RandomLabels are not guaranteed to be
processed in the same order, and may thus not always reach the threshold at the same point). |
| WeightingScheme |
Calculates the weighting of the context label. |