com.bakdata.dedupe.duplicate_detection.DuplicateDetection Maven / Gradle / Ivy
Show all versions of core Show documentation
/*
* MIT License
*
* Copyright (c) 2019 bakdata GmbH
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to deal
* in the Software without restriction, including without limitation the rights
* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
* copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in all
* copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*/
package com.bakdata.dedupe.duplicate_detection;
import com.bakdata.dedupe.clustering.Cluster;
import com.bakdata.util.StreamUtil;
import java.util.Collection;
import java.util.function.Function;
import java.util.stream.Collectors;
import java.util.stream.Stream;
import lombok.NonNull;
/**
* A duplicate detection algorithm processes a dataset of records and returns the distinct {@link Cluster}s.
* In general, all implementations will ensure that the user receives a complete clustering of the dataset, either
* through repeated updates or by providing a duplicate-free dataset.
* Consider a dataset of records A, B, A' and A, A' being duplicates. The final result will be [(A, A'), (B)]. For
* online algorithms, there may be intermediate results, such as [(A)] and [(A), (B)], which the user needs to filter
* using the cluster id.
* Now consider that we add A", which will result in a duplicate (A', A") and at the same time will cause (A, A')
* not being considered a duplicate any longer. The result of A" will be [(A', A"), (A), (B)].
* The actual implementation may use any means necessary to find duplicates and to ensure proper transitivity ((A,B)
* is duplicate and (B,C) is duplicate implies that (A,C) is duplicate).
*
* @param the type of the cluster id
* @param the type of the record
* @implSpec It is assumed that the cluster containing the new record will be the first element of the cluster list.
*/
@FunctionalInterface
public interface DuplicateDetection, T> {
/**
* Finds all duplicates in the dataset.
* Note that for online algorithms, duplicates will be repeatedly emitted with updated
* representation, since it is impossible to suppress earlier emission without blocking execution. The user needs to
* invalidate earlier results through external means (for example, putting them in a key-value store with the key
* being the id of the duplicate).
*
* @param records the records of which the duplicates should be detected.
* @return the duplicates with the above mentioned limitation for online algorithms.
*/
@NonNull Stream> detectDuplicates(@NonNull Stream records);
/**
* Finds all duplicates in the dataset.
* For online algorithms this method can only be applied on a finite stream and could be used to verify results
* in a test or compare performance to an offline algorithm.
*
* @param records the records of which the duplicates should be detected.
* @return all duplicates of the dataset.
*/
default @NonNull Collection> materializeDuplicates(final @NonNull Iterable records) {
return this.detectDuplicates(StreamUtil.stream(records))
.collect(Collectors.toMap(Cluster::getId, Function.identity()))
.values();
}
}