Natural Language Processing model upgraded to Stanford CoreNLP 4.5.1
IBM Cognos Analytics with Watson uses the Stanford CoreNLP suite as its natural language processing model in English, French, and Spanish. In Cognos Analytics 11.2.4, the CoreNLP suite was upgraded from version 3.7.0 to 4.5.1.
Suggested column Usage type and Aggregate type for the enriched data
Cognos Analytics automatically analyzes data sets and suggests the column Usage type and Aggregate type for the enriched data. Usage and Aggregate type are suggested based on calculations from multiple factors, such as name-entity recognition (column name), statistical data distribution (data sampling), and concept mapping. A change in any one of these factors can contribute to a different result.
With the upgrade to CoreNLP 4.5.1, you might eventually see Usage type suggestions that differ from previous versions. For example, you might see the Usage property attribute instead of identifier.As these are suggestions, you can customize them to the Usage type you want to use.Note: Any data that was enriched prior to the upgrade will not impact the Usage and Aggregate type.-
Suggested questions from the Assistant
When using the Assistant in Cognos Analytics, you can enter Suggest questions to receive a set of questions that you can ask. These suggestions are relevant to the context of your data source and industry.
Using the new version of CoreNLP, you might eventually receive different suggestions than in previous versions.
You can click one of the suggestions to ask that question or ask a different question.
-
Suggested interesting fields, influencers, and starting points
Cognos Analytics classifies certain table columns as interesting fields, influencers, and starting points to enable you to explore relationships in your data.
Due to the eventual different NLP classification, the user might receive different suggestions than the ones made in previous releases.