Create a brat exporter (different from brat writer)
that is thought for representing data

• create the class and arrange the code so the brat writer remains unchanged (or minimal changes)
• change the parameters to allow different improvments:
  • select the feature to represent
  • same color for each feature
  • color depending of a numerical value (between min(clear) /max(darker) )

The UDPipe segmenter is not annotating the splitted contractions due to the fix from #4. Implement the usual strategy for annotating contractions in UIMA, it is, annotating the new tokens contiguously under the span of the contraction.

DoR: #4
DoD: The segmenter annotates the components of the contractions contiguously. If the components doesn't fit, throws an exception.

• Submit issues for changes to be integrates in the DKPro
• Create an internal version with the changes over an stable version while changes are not integrated into the DKPro
• Create independent modules for changes not to be integrated in DKPro

In UDPipe 2, the words/tokens are splitted in two different lists, words list for single tokens, included the tokens resulting from separating contractions but not the contractions themselves, which are stored into multiwordTokens list.

The 1.12.x version of the UDPipe Segmenter does not seem to take this into account and this seems to be causing random errors (inexplicably, sometimes it seems to work...). Adapt the UDPipeSegmenter according to the described list setup.

DoR: Already ready.
DoD: UDPipeSegmenter works with any contractions in the text.