composite

BaseCompositeRelationalClusteringBlocker

Bases: BaseCompositeUniqueNameBlocker

Source code in klinker/blockers/composite.py

class BaseCompositeRelationalClusteringBlocker(BaseCompositeUniqueNameBlocker):
    _relation_blocker: AttributeClusteringTokenBlocker
    _rel_blocker_cls = AttributeClusteringTokenBlocker

    def concat_relational_info(
        self,
        left: KlinkerFrame,
        right: KlinkerFrame,
        left_rel: KlinkerFrame,
        right_rel: KlinkerFrame,
    ) -> Tuple[SeriesType, SeriesType]:
        """Concatenate neighbor entity attribute values with own.

        Args:
        ----
          left: KlinkerFrame: Frame with attribute info of left dataset.
          right: KlinkerFrame: Frame with attribute info of right dataset.
          left_rel: KlinkerFrame: Relation triples of left dataset.
          right_rel: KlinkerFrame: Relation triples of right dataset.

        Returns:
        -------
            (left_conc, right_conc) Concatenated entity attribute values for left and right
        """
        left_conc = concat_neighbor_attributes(
            left,
            left_rel,
            include_own_attributes=False,
            top_n_a=self.top_n_a,
            top_n_r=self.top_n_r,
            do_not_concat_values=True,
        )
        right_conc = concat_neighbor_attributes(
            right,
            right_rel,
            include_own_attributes=False,
            top_n_a=self.top_n_a,
            top_n_r=self.top_n_r,
            do_not_concat_values=True,
        )
        return left_conc, right_conc

    def _compute_rel_blocks(
        self, left, right, left_rel, right_rel, unique_blocks
    ) -> KlinkerBlockManager:
        left_conc, right_conc = self.concat_relational_info(
            left=left, right=right, left_rel=left_rel, right_rel=right_rel
        )
        left_conc = left_conc.drop_duplicates()
        right_conc = right_conc.drop_duplicates()
        left_filtered = left_conc
        right_filtered = right_conc
        if self.use_unique_name:
            left_filtered = filter_with_unique(
                left_conc, unique_blocks.blocks[left.table_name]
            )
            right_filtered = filter_with_unique(
                right_conc, unique_blocks.blocks[right.table_name]
            )
        return self._relation_blocker.assign(left=left_filtered, right=right_filtered)

concat_relational_info(left, right, left_rel, right_rel)

Concatenate neighbor entity attribute values with own.

---

left: KlinkerFrame: Frame with attribute info of left dataset. right: KlinkerFrame: Frame with attribute info of right dataset. left_rel: KlinkerFrame: Relation triples of left dataset. right_rel: KlinkerFrame: Relation triples of right dataset.

---

(left_conc, right_conc) Concatenated entity attribute values for left and right

Source code in klinker/blockers/composite.py

def concat_relational_info(
    self,
    left: KlinkerFrame,
    right: KlinkerFrame,
    left_rel: KlinkerFrame,
    right_rel: KlinkerFrame,
) -> Tuple[SeriesType, SeriesType]:
    """Concatenate neighbor entity attribute values with own.

    Args:
    ----
      left: KlinkerFrame: Frame with attribute info of left dataset.
      right: KlinkerFrame: Frame with attribute info of right dataset.
      left_rel: KlinkerFrame: Relation triples of left dataset.
      right_rel: KlinkerFrame: Relation triples of right dataset.

    Returns:
    -------
        (left_conc, right_conc) Concatenated entity attribute values for left and right
    """
    left_conc = concat_neighbor_attributes(
        left,
        left_rel,
        include_own_attributes=False,
        top_n_a=self.top_n_a,
        top_n_r=self.top_n_r,
        do_not_concat_values=True,
    )
    right_conc = concat_neighbor_attributes(
        right,
        right_rel,
        include_own_attributes=False,
        top_n_a=self.top_n_a,
        top_n_r=self.top_n_r,
        do_not_concat_values=True,
    )
    return left_conc, right_conc