What's new and changed in Watson Discovery

Installing or upgrading Watson Discovery

Cloud Pak for Data Version 4.8.5

A new version of Watson Discovery was released in April 2024 with Cloud Pak for Data 4.8.5.

Operand version: 4.8.5

This release includes the following changes:

Version 4.8.5 of the Watson Discovery service includes various fixes.

Cloud Pak for Data Version 4.8.4

A new version of Watson Discovery was released in March 2024 with Cloud Pak for Data 4.8.4.

Operand version: 4.8.4

This release includes the following changes:

Cloud Pak for Data Version 4.8.3

A new version of Watson Discovery was released in February 2024 with Cloud Pak for Data 4.8.3.

Operand version: 4.8.3

This release includes the following changes:

Version 4.8.3 of the Watson Discovery service includes various fixes.

Cloud Pak for Data Version 4.8.2

A new version of Watson Discovery was released in January 2024 with Cloud Pak for Data 4.8.2.

Operand version: 4.8.2

This release includes the following changes:

Version 4.8.2 of the Watson Discovery service includes various fixes.

Cloud Pak for Data Version 4.8.0

A new version of Watson Discovery was released in November 2023 with Cloud Pak for Data 4.8.0.

Operand version: 4.8.0

This release includes the following changes:

New features

Version 4.8.0 of the Watson Discovery service includes the following features and updates:

Enrichments are highlighted in distinct colors

When you select multiple enrichments in the advanced document view for PDF and text, each enrichment type is highlighted in the document with distinct colors. Overlapping enrichments are also highlighted in a distinct color.

Narrow the field of a search

Now, when you query data from the web client, you can specify the fields to search in. Searching specific fields improves search results when content is not indexed in the default fields. For example, content might not be indexed in the default fields when you ingest structured files or when you apply a Smart Document Understanding model. For more information, see Excerpt unavailable in the Watson Discovery documentation.

Start or stop crawling a collection

You now have the option to start or stop the scheduled crawl of a collection without changing the crawl schedule. For example, you might want to stop a crawl if you don't want the crawl to occur during a long-running task. For more information, see Stopping a crawl in the Watson Discovery documentation.

Updated tokenizer for all languages

The updated tokenizer might affect the ranking order of results for certain queries. If you observe any ranking differences in your query results, you can re-index the documents in the collection.

Watson Discovery tokenizes words when it ingests and stores data in the index, and at run time, when it analyzes queries that are submitted by users. By re-indexing the collection, you ensure that your documents are indexed with the same tokenizer that is used for matching queries.

To re-index documents, open the Manage collection page, choose a collection, and navigate to the Enrichments tab. Select a field to enrich, and then clear the field. Next, click Apply changes and reprocess and wait for the documents in the collection to be reprocessed.

For more information on re-indexing a collection, see Manage enrichments in the Watson Discovery documentation.

The part of speech enrichment is now needed only for content mining project types.

The part of speech enrichment was used for dictionary suggestion. The dictionary suggestion is now updated to work without the part of speech enrichment applied. However, for content mining projects, the part of speech enrichment is available as before.

Search for terms to label as entity examples

You can now search for terms that you want to label as entity examples in a document. You can also search for labeled and unlabeled entity examples, and you can correct any labeling inconsistencies. For more information, see Searching for examples using keywords in the Watson Discovery documentation.

Export labeled data from entity extraction

You can export labeled data that you identified when you labeled documents for entity extraction. You can use the labeled data to train or build large language models (LLMs) with services like Watson Studio and Watson Natural Language Processing (NLP). For more information, see Exporting labeled data for an entity extractor in the Watson Discovery documentation.