# Creating a clean data set from the suspect pairs labeled by tMatchPredict and the unique rows computed by tMatchPairing

This scenario applies only to Talend Platform products with Big Data and Talend Data Fabric.

In this example, there are two sources of input data:

* The suspect records labeled as duplicates and grouped by tMatchPredict.

  For an example of how to label suspect pairs with assigned labels, see [Labeling suspect pairs with assigned labels](https://help.qlik.com/talend/en-US/data-matching/7.3/labeling-suspect-pairs-with-assigned-labels).

  You can find an example of how to label suspect pairs with assigned labels on Talend Help Center ([https://help.talend.com](https://help.talend.com/)).

* The unique rows computed by tMatchPairing.

  For examples of how to compute unique rows from source data, see [Computing suspect pairs and suspect sample from source data](https://help.qlik.com/talend/en-US/data-matching/7.3/computing-suspect-pairs-suspect-sample-from-source-data) and [Computing suspect pairs and writing a sample in](https://help.qlik.com/talend/en-US/data-matching/7.3/computing-suspect-pairs-writing-sample-in-stewardship).

  You can find examples of how to compute unique rows from source data on Talend Help Center ([https://help.talend.com](https://help.talend.com/)).

The use case described here uses two subJobs:

* In the first subJob, tRuleSurvivorship processes the records labeled as duplicates and grouped by tMatchPredict, to create one single representation of each duplicates group.

* In the second subJob, tUnite merges the survivors and the unique rows to create a clean and deduplicated data set to be used with the tMatchIndex component.

The output file contains clean and deduplicated data. You can index this reference data set in ElasticSearch using the tMatchIndex component.

### In this section

* [Setting up the Job](https://help.qlik.com/talend/en-US/components/7.3/deduplication/creating-clean-data-set-from-suspect-pairs-labeled-setting-up-job)
* [Selecting the Spark mode](https://help.qlik.com/talend/en-US/components/7.3/deduplication/selecting-spark-mode-spark-batch?id=1292)
* [Configuring the connection to the file system to be used by Spark](https://help.qlik.com/talend/en-US/components/7.3/deduplication/configuring-connection-to-file-system-to-be-used-by-spark?id=1293)
* [Creating survivors from the suspect pairs labeled by tMatchPredict](https://help.qlik.com/talend/en-US/components/7.3/deduplication/creating-survivors-from-suspect-pairs-labeled-by-tmatchpredict)
* [Merging the survivors with the unique rows](https://help.qlik.com/talend/en-US/components/7.3/deduplication/merging-survivors-with-unique-rows)
* [Executing the Job](https://help.qlik.com/talend/en-US/components/7.3/deduplication/creating-clean-data-set-from-suspect-pairs-labeled-executing-job)