Skip to main content


         This documentation site is for previous versions. Visit our new documentation site for current releases.      
 

Definition class of text analytics Decision Data rules

Updated on July 5, 2022

Use the following automaticallygenerated frameworks for Decision Data rules to quickly create or update your text analytics resources, such as models and lexicons.

When creating Decision Data rules that contain the resources that are used in text analytics, you can use an updated, automatically generated framework for each class of decision data that is supported in text analysis. With the automatically generated framework for each type of text analytics Decision Data rule, you can quickly create or update the records, without having to edit the default Decision Data rule form fields.

You can create Decision Data rules that contain the following types of text analysis resources:

  • Entity extraction models
  • Entity extraction rules
  • Intent models
  • Classification models
  • Sentiment models
  • Sentiment lexicons
Note: You can create Decision Data rules in Dev Studio by clicking + CreateDecisionDecision Data. You select the definition class on the Create form of the new rule. The definition class contains the definition properties that are characteristic of each Decision Data rule type that you create.

Entity extraction models

Create decision data that contains entity extraction rules in the Data-NLP-EntityModels definition class. Entity extraction models detect entities whose names are not limited to certain patterns or dictionaries. Entity extraction models can detect names of organizations, brands, people, and so on. See the following figure for reference:

Decision Data rule that contains an entity extraction model
Configuration of a Decision Data rule with an entity extraction model.

Entity extraction rules

Create decision data that contains entity extraction rules in the Data-NLP-Rule definition class. Entity extraction rules detect entities whose names match a certain pattern or are part of a dictionary. Entity extraction rules can detect names of hardware or software products, identification numbers, emails, dates, and so on. You upload entity extraction rules as Apache Ruta scripts. See the following figure for reference:

Decision Data rule that contains an Apache Ruta script with an entity extraction rule
Configuration of a Decision Data rule with an entity extraction rule using the Apache Ruta script.

Intent analysis models

Create decision data that contains intent analysis models as part of the Data-NLP-Intent definition class. Intent analysis models determine whether the content (social media posts, comments, tweets, emails, instant messages, and so on) that you analyzed in your application was produced with an underlying intention. See the following figure for reference:

Decision Data rule that contains an intent analysis model
Configuration of a Decision Data rule with an intent analysis model.
Note: In the Intent config field, you can upload a .csv, .xls, or .xlsx file that contains your configuration for intent detection. For more information, access the configuration file that is part of the default pzDefaultIntentModel rule.

Sentiment analysis models

Create decision data that contains sentiment analysis models in the Data-NLP-SentimentModels definition class. Sentiment analysis models determine whether the analyzed text expresses a negative, positive, or neutral opinion.

As part of creating or updating the rule, you can build sentiment analysis models by using a wizard. When you update an existing model for a specific language, that model is replaced by the newly created model when the creation process finishes. In a single Decision Data rule, you can have one model for each supported language.

Tip: You can test the sentiment detection of the model that you created, based on the models that are part of the Decision Data rule.

See the following figure for reference:

Creating sentiment analysis models in a Decision Data rule
Configuration of sentiment analysis models in a Decision Data rule.

For more information about the wizard for sentiment model creation, see Determining the emotional tone of text.

Classification analysis models

Create decision data that contains sentiment analysis models and taxonomies for rule-based classification analysis in the Data-NLP-Taxonomy definition class. Either by creating a classification analysis model or by using rule-based classification analysis based on taxonomies, you can determine the categories to which a text unit should be assigned. As in the case of sentiment analysis, you can use a Decision Data rule to build classification analysis models. Additionally, you can upload .csv, .xls, or .xlsx files that contain taxonomies for rule-based classification analysis.

Tip: You can test how categories are assigned, based on the classification models and taxonomies that are part of the Decision Data rule.
Note: Business scientists can use the Decision Analytics workspace to build Decision Data rules that belong to Data-NLP-Taxonomy definition class.
Decision Data rule that contains a classification model and a taxonomy for rule-based classification
Configuration of a Decision Data rule with a classification model and a taxonomy for rule-based classification.

Sentiment lexicons

Create decision data that contains sentiment lexicons in the Data-NLP-SentimentLexicon definition class. Sentiment lexicons contain words or phrases that are assigned a sentiment value.

Decision Data rule that contains a sentiment lexicon
Configuration of a Decision Data rule with a sentiment lexicon.
  • Previous topic Best practices for providing feedback for text extraction models
  • Next topic Intelligent interaction in text analytics

Have a question? Get answers now.

Visit the Support Center to ask questions, engage in discussions, share ideas, and help others.

Did you find this content helpful?

Want to help us improve this content?

We'd prefer it if you saw us at our best.

Pega.com is not optimized for Internet Explorer. For the optimal experience, please use:

Close Deprecation Notice
Contact us