Samples for Detecting Language, Sentiment, and Actionability

The installation of eServices Manager includes samples that detect

• Language
• Sentiment
• Actionability

Screening Rules

For eServices Manager

Genesys supplies sample screening rules that analyze interactions for:

• The sentiment expressed—Positive, negative, or neutral.
• Actionability—Whether the interaction calls for attention from an agent.

To use this sample, import the file SentimentAndActionabilityScreeningRules.kme, which is located in the KnowledgeManagerHome\SentimentModel directory.

This file provides examples of screening rules for detecting sentiment and actionability, plus category trees containing categories that are assigned to any interactions that match the rules.

Models

For Content Analyzer

Language Detection

As part of the installation of eServices Manager, Genesys provides a model that classifies e-mails as English, French, German, Italian, Portuguese, Russian, or Spanish.

To import this model and its training object,

1. Open eServices Manager and select unknown as the language. If there is no such language you must create one in Genesys Administrator or Configuration Manager (under Business Attributes > Languages).
2. Then import the file lang.kme, located in <KnowledgeManagerHome>\LanguageModel. <KnowledgeManagerHome> is normally C:\Program Files\GCTI\eServices 8.5\Knowledge Manager.

The training object consists of seven categories, one for each language. Each category contains a number of text objects in its language. You can add more text objects to these categories as well. This could be especially valuable if you have a collection of text objects (such as e-mails) whose subject matter relates to your business. After you add text objects, you must train a new model to take advantage of the added data.

You can also add other languages to the model, as follows:

1. eServices Manager: On the Standard Responses tab (still with unknown selected as the language), add a category for the new language to the LanguageDetection category tree.
2. Content Analyzer:
   1. On the Training tab, select the LanguageDetection training object, select the new language category in the training object, then add text objects in the language.
   2. Train a new model that includes the new language.

You can do this for any language supported by E-mail Server (E-mail Server supports all languages that are supported by the version of JRE that is supplied with Genesys eServices). However, Genesys has not tested any language other than those listed above.

Actionability

To use the actionability sample, import the file Actionability.kme, which is located in the <KnowledgeManagerHome>\ActionabilityModel directory.

This provides:

• A model Actionability for analyzing actionability.
• The training object Actionability that created that model.
• A category tree Actionability that contains the categories to assign to interactions as a result of the analysis.

Sentiment

To deploy the sentiment sample, use the following procedure.

1. In Configuration Manager or Genesys Administrator, create a language called English_Sentiment.
2. With Knowledge Manager set to that language, import the file EnglishSentiment.kme, which is located in the <KnowledgeManagerHome>\SentimentModel directory.

This provides:

• A model SentimentSampleModel for analyzing sentiment.
  
• The training object Sentiment that created that model.
  
• A category tree SentimentDetection that contains the categories to assign to interactions as a result of the analysis.