From Unstructured Text to Valuable Insight
Organizations today generate large volumes of unstructured business text, including customer, employee, and supplier feedback, as well as operational notes. However, much of this data remains underused.
Typical sources include:
- Customer surveys, support tickets, and emails
- Sales order notes, invoice comments, and dispute descriptions
- Field logs, warehouse notes, and production comments
Individually, these data points provide useful signals. However, without process context, they are difficult to translate into action.
AI-assisted context analyzer in SAP Signavio Process Intelligence connects these data sources to process execution.
It enables organizations to:
- Understand what customers or employees are saying
- Link feedback directly to process steps
- Identify root causes of inefficiencies and experience gaps
With capabilities such as sentiment analysis and text-to-event matching, organizations can transform unstructured text into actionable, process-driven insights.
Why AI-Assisted Context Analyzer Matters
AI-assisted context analyzer enables a new layer of analysis where business text becomes part of process intelligence instead of being reviewed in isolation.
Key Value
- Identify root causes hidden in text data
- Correlate process deviations with real explanations
- Reduce manual investigation effort
- Improve both customer experience and operational performance
Core Capabilities
1) Sentiment Analysis
Automatically classifies free text into: positive Neutral, or negative to help identify where customer or employee experience requires attention.
2) Text-to-Event Matching
Maps unstructured text to process activities and enables analysis such as:
- Which process steps generate the most complaints
- Where operational issues occur most frequently
Industry Use Cases
Detailed Use Cases
Note
AI-assisted context analyzers only provide analytical decision support. You can then make decisions and trigger actions based on this support.
Prerequisites
Access Requirements
Data Requirements
- Data view or source data with: Key field column and Free text column
- Each case is linked to one free text entry
- For cases with multiple texts, only the first text will be analyzed and matched
Pipeline Setup
- Case-based data pipeline
- Business object and event collector configured and enabled
- Extraction, transformation, and load (ETL) has successfully run for the pipeline
- Event log data is loaded into analysis configuration
How to Set Up AI-Assisted Context Analyzer
Step 1: Connect Free Text Data
Load customer or operational text into SAP Signavio Process Intelligence and ensure it is part of a case-based pipeline.
Step 2: Create Context Analyzer
Go to the Context Analyzers tab in your pipeline and create a new context analyzer.
Step 3: Choose the Task
Select the task you want to configure:
- Sentiment analysis: Classifies each free text as positive, neutral, or negative
- Text-to-event matching: Recommends matches between free text and process events
Step 4: Select Input Data
For your input data, choose either from:
Option A: Existing Data View
Select your data view and choose the free text and key field columns
Option B: Existing Source Data
Select your source data and table, then choose the free text and key field columns.
Review the selected input data in the details card.
Step 5: Configure
Common setting for both tasks:
- Anonymization is enabled by default and masks personally identifiable information in the input text (see Data Masking).
Additional setting for text-to-event matching:
- Overlay sentiment label in widget output is available and enabled by default. This displays sentiment scores and color indicators in the Process Discovery widget.
While both tasks share the same setup flow, their outputs differ in how insights are visualized and analyzed, as shown in the next sections.
Once the setup is complete, click Create.
Step 6: Run
To run context analyzers, you can:
- Trigger the pipeline run with Run ETL or Run T&L manually
- Wait for the next scheduled run
Track progress in Event Log Load step in the Pipeline Log.
Once the run is complete, you can find the Context Analyzer table with details such as duration, message and execution ID (use Show IDs toggle to display) in Event Log Load step.
Next, go to the Process Data Model tab to view the output attributes for each context analyzer.
Analyzing Results Using Outputs
Sentiment analysis
- Sentiment label attribute (case-level)
- Free text attribute (case-level)
Text-to-event matching
- Free text match count metric
- Free text attribute (event-level)
- If sentiment overlay is enabled:
- Sentiment score metric (0-100 range)
- Sentiment label attribute (event-level)
Step 1: Create a Dashboard
Access your linked analysis configuration by clicking on the card with blue highlight in the Overview tab.
Create a customizable dashboard.
Step 2: Sentiment Analysis Visualization
Create a Charts and Tables widget.
In widget builder, add the following attributes to dimensions:
- Case
- Sentiment label (case-level)
- Feedback text (case-level)
Step 3: Text-to-Event Visualization
Create a Process Discovery widget.
Select “Text-to-event matches” metric in metric dropdown. You will see the count of free-text records matched to each event.
Select “Text-to-event sentiment score” in metric dropdown. You will see the sentiment-colored overlay on each event along with its score.
- A positive sentiment is >70 and has a green overlay
- A neutral sentiment is in the range of 50-70 and has an orange overlay
- A negative sentiment is <50 and has a red overlay
These ranges and color overlays can be customized (see Progression in Metric Settings).
Once widgets are configured, make sure to save your dashboard to keep your progress.
With AI-assisted context analyzer, unstructured text becomes part of your process analysis.
By combining sentiment analysis and text-to-event matching, you can enrich your event log with additional attributes and metrics derived from business text. These insights can be analyzed in dashboards to better understand how feedback and operational notes relate to process behavior.
Once configured, the analyzer runs as part of your data pipeline, enabling continuous analysis without additional manual effort.
