Nuix NLP v1.2

The following new features and fixed issues are included in Nuix NLP.

New Features

Shared Keycloak

NLP now integrates with Keycloak for Neo-shared user accounts and configuration. This allows for a more cohesive platform with a shared user account across components.

Compound Lexeme References

When using the NLP model builder, users can create complex extractions with an improved user experience and protection from breaking existing compound lexemes by viewing referenced files.

Compound Lexeme Exclude and Include

In compound lexemes, you can now add ‘exclude’ or ‘include’ to any items and reduce false positives from space entities. This allows additional support for more granular extractions.

Compound Lexeme Language Support

NLP now has compound lexeme language support which provides a consistent experience for languages across the product skills, topics, and compound lexemes. The language support allows for more control over which compound lexemes are applied to processing jobs.

Users can add internal language IDs to compound lexemes. You can also add a ‘language agnostic’ flag for compound lexemes which can be applied to all languages.

NLP Model updates

NLP model updates were added to compound lexemes, enhanced compound lexemes, and enhanced topics.

New Compound Lexemes

Australian Business Number (AU)

Australian Company Number (AU)

Bank State Branch (AU)

Immicard Number (AU)

Individual Healthcare Identifier (AU)

Medicare Number (AU)

Company Registration Number (UK)

National Insurance Number (UK)

NHS Number (UK)

Sort Code (UK)

UTR Number (UK)

Enhanced Compound Lexemes

BTC Address

Person

Address

Enhanced Topics

Cryptocurrency

Social Media

Mobile & Wireless

Email & Messaging

Additional updates include:

Update to Compound Lexemes to allow simple and complex item types to both include and exclude simple or complex item types. Every simple or complex item type will have an optional “includes” or “excludes” field. This field can take in any complex or simple item type, which will then act as a check to (in the case of include) ensure that the items are within the scope of the match, or if excluded are not within the scope of the match. The scope of the include and exclude check is limited to the match of the block that it belongs to.

Update in the API to allow adding anything to the Value field for a “NamNormalization Rule” in addition to the Mapping field.

Other entity types include: { PERSON, GROUP, FACILITY, ORGANIZATION, GEO_POLITICAL_ENTITY, LOCATION, PRODUCT, EVENT, WORK_OF_ART, LAW, LANGUAGE, DATE, TIME, PERCENT, MONEY, QUANTITY, ORDINAL, CARDINAL }.

Removed “personNames” field from the search results.

Enabled pipelines and campaigns in the top-right-hand drop-down for all deployments.

Translated new user interface phrases into German and Spanish.

Fixed Issues

If a user tries to call either the /topic/delete API or /country/delete API on the dictionary service an error response is returned This is now fixed.

The user is unable to add User tags on Dictionaries and Topics. This is now fixed.

The user was unable to process a document through the Campaigns user interface. This is now fixed.

The named items for Nested Compound Lexemes do not carry through to the results. This is now fixed.

An error occurred when importing Compound Lexemes with nested Compound Lexemes if the nested Compound Lexeme is in the export file and in the export hints. This is now fixed.

The Compound Lexeme export screen tells the user that it produces a zip file when it produces a JSON file. This is now fixed.

The Risk Rules list disappears when moving between risk-rule types. This is now fixed.

A number of index out-of-range errors appeared in engine-app logs while processing a dataset. This is now fixed.

If the user exports a Compound Lexeme that contains a nested Compound Lexeme, the export file does not include the nested Compound Lexemes labels as a part of the “importHints”. This is now fixed.

When importing a Compound Lexeme with a reference to another Compound Lexeme, that already exists, an unnecessary clone is created. This is now fixed.

Hovering over a selected lexeme inside a Compound Lexeme shows (lexeme / topic / lexeme). It should show (dictionary / topic / lexeme). This is now fixed.