colrev.ops.dedupe.Dedupe

class colrev.ops.dedupe.Dedupe(*, review_manager, notify_state_transition_operation=True)

Bases: Operation

Deduplicate records (entity resolution)

Methods

apply_merges	Apply deduplication decisions
check_precondition	Check the operation precondition
conclude	Conclude the operation (stop Docker containers)
connected_components	Find the connected components in a graph.
decorate	Decorator for operations
fix_errors	Fix lists of errors
get_info	Get info on cuts (overlap of search sources) and same source merges
get_records_for_dedupe	Get (pre-processed) records for dedupe
main	rtype: Any
merge_based_on_global_ids	Merge records based on global IDs (e.g., doi)
merge_records	Merge records by ID sets
notify	Notify the review_manager about the next operation
unmerge_records	Unmerge duplicate decision of the records, as identified by their ids.

Attributes

DUPLICATES_TO_VALIDATE
NON_DUPLICATE_FILE_TXT
NON_DUPLICATE_FILE_XLSX
PREVENTED_SAME_SOURCE_MERGE_FILE
SAME_SOURCE_MERGE_FILE
debug
type

apply_merges(*, id_sets, complete_dedupe=False, preferred_masterdata_sources=None)

Apply deduplication decisions

id_sets : [[ID_1, ID_2, ID_3], …] :rtype: None

• complete_dedupe: when not all potential duplicates were considered,

we cannot set records to md_procssed for non-duplicate decisions

check_precondition()

Check the operation precondition

Return type:

None

conclude()

Conclude the operation (stop Docker containers)

Return type:

None

classmethod connected_components(id_sets)

Find the connected components in a graph.

Return type:

list

Args:

id_sets (list): A list of id sets.

Returns:

list: A list of connected components.

classmethod decorate()

Decorator for operations

Return type:

Callable

fix_errors(*, false_positives, false_negatives)

Fix lists of errors

Return type:

None

get_info()

Get info on cuts (overlap of search sources) and same source merges

Return type:

dict

classmethod get_records_for_dedupe(*, records_df, verbosity_level=0)

Get (pre-processed) records for dedupe

Return type:

DataFrame

merge_based_on_global_ids(*, apply=False)

Merge records based on global IDs (e.g., doi)

Return type:

None

merge_records(*, merge)

Merge records by ID sets

Return type:

None

notify(*, state_transition=True)

Notify the review_manager about the next operation

Return type:

None

unmerge_records(*, current_record_ids)

Unmerge duplicate decision of the records, as identified by their ids.

The current_record_ids identifies the records by their current IDs and unmerges their most recent merge in history.

Return type:

None