Run screening and view reports

Start screening

To start Transaction Screening, do the following:

1. In the AI Agents list, click the installed AI Agent (variation).

2. On the AI Agent variation's details page, open the process definition and click a Business Process (BP) instance name.

   

3. Click the name of the required BP, navigate to the Data tab, and add your input data in one of the ways:

   • Upload a CSV file.
   • Use the streaming API input option.
   • Get data from a connector.

   

4. Go to the Run tab and click Run This Process.

   

   The Business Process started successfully window appears.

5. Click View results to see the Results tab.

   

View results

HTML reports

If you configured HTML reports earlier, the reports are available in S3 at doc-upload/payment_sanctions_screening/reports/html. To review detailed screening results, complete the following steps:

1. In the Payment Sanction Screening BP, go to Results > Final results > Steps and select the Sanctions Screening JNW (generate-report-task-processor) step.

2. On the Data tab, in the links_to_report column, copy the report link and paste it in a browser window to view the report.

   

   The Screening Review report opens in a separate window.

   

3. Click the hit ID to view the details.

   

An HTML report contains the following data:

• Input fields from the screened and santioned sides as Input Information and Watch List Information respectively

• Final decision made by Tara as Decision and Narrative

• Map snapshot of the verified screened location as SCREENED Location if Geocoder is turned on and the address is verified

  

• Gold value as Actual Decision (this value might be present when running a CSV input file)

• Copy of the original JSON request as Raw JSON Input used for debugging and analysis

• Copy of the rule JSON object that led to the decision as Rules Model JSON used for debugging and analysis

A sample decision response

{
    "decision": {
        "messageId": "7023270",
        "suggestedAction": "NO_DECISION",
        "comment": "PARTIALLY_RESOLVE: 1 hit(s) escalated by [name_match]; 1 hit(s) resolved by [name_mismatch_strong]; ",
        "hitDecisions": [
            {
                "reasons": [
                    {
                        "score": 1,
                        "code": "name_match",
                        "description": "Name match. Screened name (Eddie ) matches sanctioned name (Eddie, Oliver Hane) or primary name or aliases."
                    }
                ],
                "hitId": "823185168",
                "suggestedAction": "NO_DECISION",
                "comment": "Name match. Screened name (Eddie ) matches sanctioned name (Eddie, Oliver Hane) or primary name or aliases."
            },
            {
                "reasons": [
                    {
                        "score": 0.4401,
                        "code": "name_mismatch_strong",
                        "description": "Name mismatch. Screened name (Joelle Heathcote-Howe limited liability company) does not match sanctioned name (Heathcote Cartwright-Kerluke) and aliases."
                    }
                ],
                "hitId": "603908759",
                "suggestedAction": "RESOLVE",
                "comment": "Name mismatch. Screened name (Joelle Heathcote-Howe limited liability company) does not match sanctioned name (Heathcote Cartwright-Kerluke) and aliases."
            }
        ]
    },
    "requestId": "859e93c6-726c-4934-bde6-d8b5eb4d8421",
    "statusDetails": "",
    "html_report_link": "https://instance-s3.cloud.workfusion.com/doc-upload/tara-html-reports/e383607a-5447-481e-88b6-9e913bf3af91/message-7023270/message-hits.html",
    "status": "COMPLETED"
}

QC reports

In QC reports, report data is populated at the hit level for each incoming request within configured timeframe. The data used to generate reports is stored in the pss_request_processing_v4 Data Store.

You can run reports manually or schedule them. The recommended scheduling options are as follows:

• End of business day. Select the Today option to collect data for the current day.
• After midnight. Select the Last x days option and enter 1 in the field to collect data for the previous day. For more details, see Install and configure | Output.

Generated CSV files are saved to S3 at doc-upload/payment_sanctions_screening/reports/qc. Before new reports are created, all previous reports are moved to the doc-upload/payment_sanctions_screening/reports/qc/reports_archive directory.

info

You can change the S3 directory location if you use a custom S3 bucket. The inner directory structure must remain the same.

The report template located at doc-upload/payment_sanctions_screening/reports/templates/qc_report_template.csv. When you create a custom report, place it in S3 at doc-upload/payment_sanctions_screening/reports/templates/qc_report_template_custom.csv.

The report can contain the following columns:

Column	Description
message_id	Alert ID; an alert may contain many hits.
msg_type	Message type; may contain type and subtype or subtype only.
content_format	Content format.
hit_id	Unique hit ID within the alert.
hit_tagcontent	Parsed or supplied value defined by tag_name in the payment message.
hit_tagname	Name of the tag in the message that was identified as a hit during screening.
hit_hittext	Part of the text that triggered a sanctioned match.
hit_hittype	Type of the sanctioned hit.
hit_screeneddata_type	Type of the screened entity that was identified as a hit. The value can be provided or predicted and is used in the final decision.
hit_screeneddata_name	Name of the screened entity that was identified as a hit. The value can be provided or predicted and is used in the final decision.
hit_screeneddata_address	Address of the screened entity that was identified as a hit. The value can be provided or predicted and is used in the final decision.
hit_screeneddata_dob	Date of birth of the screened entity that was identified as a hit. The value can be provided or predicted and is used in the final decision.
hit_screeneddata_country	Country of the screened entity that was identified as a hit. The value can be provided or predicted and is used in the final decision.
hit_screeneddata_id	ID of the screened entity that was identified as a hit. The value can be provided or predicted and is used in the final decision.
screen_data_source	Source of the screened data: screened, NER, or combined.
hit_sanctiondata_name	Name of the sanctioned entity that matched. This can be the best matching alias or the primary name.
hit_sanctiondata_primaryname	Primary name of the sanctioned entity that matched.
hit_sanctiondata_type	Type of the sanctioned entity that matched.
hit_sanctiondata_dob	Date of birth of the sanctioned entity that matched.
hit_sanctiondata_country	Country of the sanctioned entity that matched.
hit_sanctiondata_id	ID or code of the sanctioned entity that matched.
hit_sanctiondata_sourceentryid	Source ID of the sanctioned entity that matched.
alert_decision	Alert-level decision.
hit_decision	Hit-level decision.
hit_decision_code	Main reason for the disposition.
hit_decision_comment	Detailed explanation of all reasons for the disposition.
hit_name_matcher_score	Highest name matcher score between the screened name and all sanctioned names.
hit_type_confidence_score	NER model confidence score for the predicted screened type.
hit_ner_type_original	Original type predicted by the NER model.
is_auto_qc	Defines if the hit or alert is marked for automatic quality control.
tara_version	AI Agent version.
ner_model_version	NER model version.
decision_model_version	Decision model version.
request_id	Unique request ID. The value can be used as a link to the processing table.
last_timestamp	Timestamp when the request is completed; the alert-level timestamp.
alert_execution_time	Total record execution time in seconds at the alert level.
alert_completion_status	Processing status: COMPLETED, IN_PROGRESS, or FAILED.
gold_decision	Expected decision if available. Used for analysis and training.
gold_decision_comment	Expected decision comment if available. Used for analysis and training.

CLDS reports

The data used to generate CLDS reports is located in the pss_request_processing_v4 Data Store. If a hit matches historical data, it includes a CLDS object with all necessary report information.

You can run reports manually or schedule them. The recommended scheduling options are as follows:

• End of business day. Select the Today option to collect data for the current day.
• After midnight. Select the Last x days option and enter 1 in the field to collect data for the previous day. For more details, see Install and configure | Output.

Report templates are located at doc-upload/payment_sanctions_screening/reports/templates, for example:

• CLDS report template with standard columns: doc-upload/payment_sanctions_screening/reports/templates/clds_report_template.csv
• Custom CLDS report template: doc-upload/payment_sanctions_screening/reports/templates/clds_report_template_custom.csv

Generated CSV files are saved in S3 at doc-upload/payment_sanctions_screening/reports/clds. Before new reports are created, all previous reports are moved to the doc-upload/payment_sanctions_screening/reports/clds/reports_archive directory.

Generated filenames follow the format: CLDS-Payment-Screening-{date}_{time}-{part}.csv For example, CLDS-Payment-Screening-20250113_225716-1.csv.

The report template is located in S3 at doc-upload/payment_sanctions_screening/reports/clds/clds_report_template.csv. This template cannot be customized and includes the following predefined columns:

Column	Description
hash_id	Calculated unique hit hash value.
id	ID from a system generating an alert.
message_id	Message ID of an incoming alert.
reference_id	Content of the [REFERENCE] field of an incoming alert.
message_type	Available in a subtype field of the message. Content of the [TYPE] field of an incoming alert.
content	Content of an incoming message.
hit_id	Unique hit ID within an alert.
hit_tagcontent	Parsed or supplied value defined by the tag name in a payment message.
hit_tagname	Name of the tag in a message that was a hit during screening.
hit_hittext	Part of the text that triggered a sanctioned match.
hit_hittype	Type of a sanctioned hit.
hit_sanctiondata_name	Name of the matching sanctioned entity (best matching alias or primary name).
hit_sanctiondata_primaryname	Primary name of the matching sanctioned entity.
hit_sanctiondata_type	Type of the matching sanctioned entity.
hit_sanctiondata_sourceentry_id	Source ID of the matching sanctioned entity.
hit_sanctiondata_sourcekeywords	Keywords for sanctioned listing sources.
payment_sender_bic	Available in the payment section of a message. Value of the [SENDER] field of an incoming FUF message.
payment_receiver_bic	Available in the payment section of a message. Value of the [RECEIVER] field of an incoming FUF message.
payment_direction	Incoming or outgoing. Available in the payment section of a message.
hits_count	Number of hits in an alert.
hash_counter	Number of times each hit hash occurred.
hash_minimum_counter	Number of alerts required to clear for a hash ID before continuous learning begins auto-disposition.
hash_enable_date	Date when each hit hash achieves the CLDS-enabled status.
hash_status	Current enabled or disabled status of the hash.
alert_released	Indicates whether all hits are resolved by continuous learning (Y).
hit_decision	Indicates whether the final decision for the individual hit is RESOLVE or NO_DECISION.
hit_decision_comment	Explanation of the hit decision if the historical data contains analyst comments. For RESOLVE decisions, the field contains the most recent analyst comment).
alert_date_received	Date when a message was sent to Tara.

A hash ID is calculated as the sum of hit-level fields and message-level fields used for decision reapplication and tag content. The hit-level hash comprises predefined hard-coded fields. You can configure message tags during reapplication data setup.

Analytics Data Stores

Execution results are also recorded in the message and hit analytics Data Stores. To view the Data Stores, go to Advanced > Data Stores and select:

• pss_messages_analytics_v4 to view message-level decisions
• pss_alerts_analytics_v4 to view hit-level decisions

To analyze the results, you also need access to the original requests. All requests are saved in the request processing Data Store.

1. Go to Advanced > Data Stores and open the pss_request_processing_v4 Data Store.

   

2. Click Download, then select All Records > Comma Separated (.csv) to download all records as a CSV file.

   

   Once the file is downloaded, you can review the requests and responses to verify the decisions.

Event log

To view the event log, go to the Results tab. If any issues occur during BP execution, an exclamation mark appears.

• To view the errors, click Event log. The event log opens in a separate window.

  

• To download the report, click Export to Excel.

  

tip

For further analysis, also review the Tara: Payment sanction screening overview dashboard.