INNER CODE UNIT · Python

TextAggregatorMixin

NorskRegnesentral/skweak · skweak/aggregation.py:262

class TextAggregatorMixin(AbstractAggregator):
    """Implementation of a subset of methods from AbstractAggregator when
    the aggregation is performed for text/span classification.
    This class should not be instantiated directly."""   

    def get_observation_df(self, doc: Doc):
        """Returns a dataframe containing the observed predictions of each labelling
        sources for the document. The content of the dataframe depends on the prefixes.
        The dataframe has one row per unique spans."""

        # Extracting the sources to consider (and filtering out the ones to avoid)
        sources = self._get_sources_to_aggregate(doc)

        # Extracts a list of unique spans (with identical boundaries)
        unique_spans = set((span.start, span.end)
                           for s in sources for span in doc.spans[s])
        sorted_spans = sorted(unique_spans)
        spans_indices = {span: i for i, span in enumerate(sorted_spans)}

View source record →

📰 Research Paper
Loading…
⏳ Fetching content…