ixp
latest
false
- Introduction
- Setting up your account
- Balance
- Clusters
- Concept drift
- Coverage
- Datasets
- General fields
- Labels (predictions, confidence levels, label hierarchy, and label sentiment)
- Models
- Streams
- Model Rating
- Projects
- Precision
- Recall
- Annotated and unannotated messages
- Extraction Fields
- Sources
- Taxonomies
- Training
- True and false positive and negative predictions
- Validation
- Messages
- Access Control and Administration
- Manage sources and datasets
- Understanding the data structure and permissions
- Creating or deleting a data source in the GUI
- Uploading a CSV file into a source
- Preparing data for .CSV upload
- Creating a dataset
- Multilingual sources and datasets
- Enabling sentiment on a dataset
- Amending dataset settings
- Deleting a message
- Deleting a dataset
- Exporting a dataset
- Using Exchange integrations
- Model training and maintenance
- Understanding labels, general fields, and metadata
- Label hierarchy and best practices
- Comparing analytics and automation use cases
- Turning your objectives into labels
- Overview of the model training process
- Generative Annotation
- Dastaset status
- Model training and annotating best practice
- Training with label sentiment analysis enabled
- Training chat and calls data
- Understanding data requirements
- Train
- Introduction to Refine
- Precision and recall explained
- Precision and Recall
- How validation works
- Understanding and improving model performance
- Reasons for label low average precision
- Training using Check label and Missed label
- Training using Teach label (Refine)
- Training using Search (Refine)
- Understanding and increasing coverage
- Improving Balance and using Rebalance
- When to stop training your model
- Using general fields
- Generative extraction
- Using analytics and monitoring
- Automations and Communications Mining™
- Developer
- Exchange Integration with Azure service user
- Exchange Integration with Azure Application Authentication
- Exchange Integration with Azure Application Authentication and Graph
- Fetching data for Tableau with Python
- Elasticsearch integration
- Self-hosted Exchange integration
- UiPath® Automation Framework
- UiPath® Marketplace activities
- UiPath® official activities
- How machines learn to understand words: a guide to embeddings in NLP
- Prompt-based learning with Transformers
- Efficient Transformers II: knowledge distillation & fine-tuning
- Efficient Transformers I: attention mechanisms
- Deep hierarchical unsupervised intent modelling: getting value without training data
- Fixing annotating bias with Communications Mining™
- Active learning: better ML models in less time
- It's all in the numbers - assessing model performance with metrics
- Why model validation is important
- Comparing Communications Mining™ and Google AutoML for conversational data intelligence
- Licensing
- FAQs and more
Communications Mining user guide
Last updated Aug 11, 2025
Each use case will fall into one of the following categories, based on the intended
outcomes, that is, the objectives:
- Analytics and monitoring.
- Automation.
- Sometimes, even both.
These intended outcomes dictate how you annotate your data, and structure your taxonomies.
The way you structure and train your model can vary significantly depending on your objective. For example, obtaining detailed analytics for a communication channel compared to auto-routing inbound requests into different workflow queues.
Before building a taxonomy to meet either analytics or automation focused objectives, make sure you understand the differences between them:
Analytics use cases
- More extensive taxonomies with higher numbers of labels, usually, between 50 and 150.
- Fewer pinned examples per child label, usually, between 25 and 75.
- The main objective is to achieve detailed coverage across a broad range of topics to identify areas for improvement.
- Example use cases: opportunity, discovery, and the voice of the customer.
Automation use cases
- Smaller taxonomies with lower numbers of labels, usually, between 20 and 60.
- Higher pinned examples per child label, usually, between 50 and 100, or more.
- The main objective is to achieve high precision and recall for all automation labels to maximize accuracy and minimize exceptions.
- Example use cases: auto-routing and query management
Objectives
- Focus on gaining a detailed understanding of the various processes, issues, and sentiments within one or more communication channels.
- Provide initial insights once the model is trained, and an ongoing ability to monitor changes and trends within the dataset over time.
- Continuously help to identify, quantify, and prioritize opportunities to make improvements within the communications channel, whether to improve efficiency, customer experience, or control.
- Reduce the risk of not delivering expected ROI of change investment by effectively quantifying opportunities.
Examples
- Accurately identify the most valuable change opportunities, driving tighter ROI for specific initiatives and reduce risk of not delivering expected benefits.
- Improves customer satisfaction and service quality by identifying and driving impactful improvements in products and services.
- Reduces client-impacting issues and internal cost-to-serve.
- Better target potential customers and enable proactive customer retention by measuring CLTV drivers.
- Increase visibility and control of risks hidden in communication channels through monitoring and alerts, ensuring participants receive data they need when they need it and enable proactive remediation.
- Provide quality assurance across customer support teams, monitoring effective agent resolution.
- Empower managers to address performance issues proactively.
Labelling
- Given their purpose, they have detailed, extensive taxonomies.
- Despite higher numbers of labels, they usually have fewer pinned examples per label than automation focused datasets.
- As they are intended to capture more specific labels across an entire dataset, they sacrifice a bit of accuracy in their predictions in order to achieve detailed coverage across a broad range of topics.
Objectives
- Make efficiency gains, free up FTE capacity for value-add work, and improve CX by reducing processing times and error rates.
- Bring control, visibility, and standardization to processes.
Examples
- Reduce FTE effort by 5-10% through auto-triaging.
- Reduce turnaround time for automated tasks by 100%.
- Eliminate process issues due to incorrect classification, prioritisation, and misrouting.
- Eliminate capacity constraints and volume sensitivity.
- Enable expansion to end-to-end automation of processes or queries.
- Reduce risk around business processes through increased controls.
- Improve client satisfaction, such as CSAT or NPS, and service quality through reduced process latency.
Labelling
- These have small taxonomies with higher numbers of pinned examples for every label.
- More examples are needed per label to ensure high precision and recall, and to catch various edge cases in the dataset.
- Each label involved in an automation should seek to maximize precision and recall, although it is not usually possible for both precision and recall to reach 100%. Depending on the use case, you might optimize one slightly over the other. There will almost always be some exceptions, so make sure you have a proper exception process in place for any automation use case.
Note:
Datasets trained for automation objectives can still deliver valuable analytical insights, though they may lack the granularity of those designed to answer more detailed questions.
For more details on how to turn your objectives into labels and an appropriate taxonomy, whether for analytical or automation purposes, check Turning your objectives into labels.
