Skip to main content
Version: 2.9

F.A.Q.

Frequently asked questions​

Q: Actual names appear in the Keyword list in the Negative News Review UI, such as 'Paul.' Why is this even considered, especially when the name searched does not include this name?

A: The set of keywords is determined dynamically per each news article. When article text is analyzed, the keywords and phrases most likely to influence the decision are highlighted. The process does not distinguish common words vs. names and may highlight both. For example, after a sufficient number of training examples, it is very likely to start highlighting "Donald" as a word the user needs to pay attention to for review.

Q: Can one dictate the language they would like to see in the report?

A: Yes, with customization. Both Google and Factiva allow you to specify the language of the results. This Effectively filters out all articles that are not in a selected language. Please note that the connectors' source code may need to be modified to pass the additional language parameter.

Q: Can the solution automatically translate the articles and present them in the language specified by the user?

A: Yes. The AI Digital Worker comes with out-of-the-box Google Translate connector. As an extra step, the connector can be used to translate articles into English and process them through the rest of the BP.

Q: Can one modify the output report?

A: Yes, the report is fully customizable. Additional Reporting resources can be provided for further details.

Q: Can one create a weekly report of new Tagged and UnTagged Keywords?

A: The recommended approach to creating such a list is through the custom analytics dashboard published to Control Tower. Additional resources can be provided for further details.

Q: How is the search query for news articles constructed?

A: The search is completed using rules-based automation. For example, in the case of Google News used as a news article source: the RPA functionality of WorkFusion is used to launch a browser, navigate to Google News, and perform the search in the same way a person would. The search is configured as follows:

Q: News articles displayed in WorkFusion Negative News are not the same as the search results retrieved manually. Why is that?

A: There are two common reasons a comparison of searches (manual vs. WorkFusion) may produce different results:

  • It is not the same search; you are conducting searches with other parameters.
  • When looking at the results of WorkFusion's analysis including articles ranked by relevence for the Negative News Analysis, the order will be different.

Q: Can similar articles be grouped?

A: Yes. Both Factiva and Google News support this option, and we have it turned on by default. Similar articles will come as one for the analysis.

Q: How the solution addresses Audit and Regulation concerns? Does it explain the decisions made?

A: The Negative News skill comes with several features that address the regulator's need for a transparent and documented explanation of decisions. Also, this process is not unattended. It keeps a 'human-in-the-loop' who reviews WorkFusion's analysis. This leaves a clear identifiable human responsibility on top of the expeditious search and research supported by WorkFusion.

WorkFusion enables and helps to mitigate the risk that adverse media monitoring is subject to, where the institution's SME or analysts decide the final material or immaterial risk. WorkFusion does not determine, evaluate, or exclude risk from the FI, rather it enables better risk management.

Also, WorkFusion is automating the analyst's work and saving them time in three ways:

  • Ingesting news more quickly
  • Conducting initial news analysis, news ranking, and news summaries to help the analyst conclude more quickly
  • Generating an output report based on the information, WorkFusion's investigation and any inputs of the analyst

So, while this can save 50-80% of an analyst's time, the actual decision remains human-led.

Finally, the analysis produced by WorkFusion comes with a complete audit trail and a detailed output report. This should give more transparency and documentation than a decision made manually, which, by contrast, might only highlight and record critical points rather than a complete set of data and analysis.

Q: What news sources does the solution use?

A: Connectors for the following news sources are available out of the box:

  • Factiva
  • Google
  • Google News
  • LexisNexis

Q: Can one add a different news article source?

A: Changing news sources/search engine or adding additional ones is a relatively simple configuration to implement. As a rule of thumb, it can take two weeks of development for a completely new source. A key element of configuring the process for production requirements includes ensuring the default search parameters are correct, matching your business requirements.

Known issues​

Manual Task couldn't display options for High-Risk Countries

Technical details: Import MT from 9.x to 10.x with the following parameters:

  • A "select one" answer

  • Data Store linked to this answer.

    Instead of the Data Store name in DB, it saves options from Data Store. The issue is fixed in IA Cloud v10.1.3 and later.

To fix the issue manually, do the following:

  1. Go to the imported Manual Task (inside BP or Manual Tasks → View All → Negative News Review) and open it in Design mode.
  2. Open Select One answer with Data Store (High-Risk Countries).
  3. Only click Save in the Answer box, don't perform any other changes.
  4. Click Save the Manual Task.
BEP worker fails upon termination due to java.lang.OutOfMemoryError: Metaspace

Under heavy load, this kind of error may appear in BEP workers' logs and cause the workers to restart.

To mitigate the issue, the workers' metaspace memory must be increased to 256 MB.

# On Master server
vi /opt/workfusion/vds-data/workers/app/com.workfusion.spa.ct/worker-app/10.1.0.31/worker.yml
# change "metaspace: 128" to "metaspace: 256"
Last updated on