TABLE OF CONTENTS
The list below describes key steps required to set up and deliver a Communications Mining use case:
1. Accessing Communications Mining
Automation Cloud users
If you are an Automation Cloud user and have AI units enabled, Communications Mining can be accessed via the Automation Cloud. If you don't have any AI units but want to start using Communications Mining, please contact your account manager.
For further information on how to access Communications Mining on Automation Cloud for the first time, click here.
For further information on how to manage your account on Automation Cloud, click here.
You don't need to be an Automation Cloud user to access Communications Mining.
For further information on how to access Communications Mining for the first time, click here.
For further information on how to manage your account, click here.
2. Creating a project
Projects can be thought of as restricted workspaces. Each dataset and data source is associated with a specific project, with users requiring permissions in those projects to be able to work with the data within them. Datasets in one project can be made up of data sources multiple projects. Users will just require permissions in both projects to view and label the data.
For more information on data structure, click here.
For Automation Cloud users, every tenant has a 'Default Project' that all users within the tenant have access to. Before uploading data, creating datasets and training models, it's strongly recommended to create a new project with access limited to only those individuals who require access to that data. Once created, it's difficult to move data sources and datasets into different projects.
To create a new project, follow the steps here.
3. Adding users to a project with correct permissions
Access to Communications Mining tenants, projects, data sources and datasets is controlled by strict user permissions. Permissions need to be allocated per-user. They can provide access to sensitive data and allow users to perform a range of different actions in the platform. Users should only be given permissions they need to fulfil their roles. See here for a more detailed explanation of user permissions.
To create a new legacy user, follow the steps here.
To add a user to a project, follow the steps here.
4. Creating a data source
Data sources are collections of raw unlabelled communications data of a similar type (e.g. emails from a shared mailbox or a collection of NPS survey responses).
Creating a source in the GUI essentially sets up an empty source with defined properties, that data can then be uploaded to via the API. The setup of this source can also be done via the API.
Once the source is created, data can be uploaded via:
- Integration (i.e. Exchange integration, Salesforce integration, etc.)
- Static CSV upload
To create a new data source in the GUI, follow the steps here.
To upload a CSV file into a source, follow the steps here.
For integration guidance and technical documentation, click here.
5. Creating a dataset
Datasets are comprised of 1 or more data sources (max 20) and the model that you train.
Please note that
To create a new dataset, follow the steps here.
For more information on using multilingual datasets and sources, click here.
6. Training and maintaining a model
Prerequisites before you start training a Communications Mining model include:
- Defined objectives and success criteria
- Designed taxonomy of labels and entities
- Business SMEs with domain-specific knowledge
- Ring-fenced time to train the model
Model training process consist of 3 key phases: Discover, Explore, and Refine. Our new feature 'Train' provides a guided training experience that walks users through each phase of training step-by-step.
Any model that is being used in production needs to be effectively maintained to ensure continued high-performance. This includes a) preventing concept drift, and b) creating an exceptions process.
For further model training information, see the links below:
7. Exploring analytics
The platform has built-in reporting and analytics capability that can help you identify potential issues and improvement opportunities across your communications channels, for example:
- Requests that are transactional in nature can be good candidates for automation or self-service
- Requests that get no response or follow-up can potentially be eliminated
- No-action required emails (i.e. OOO, spam, auto-generated emails, thank you emails) can potentially be deleted from a mailbox
- Urgent queries that need to be prioritised and resolved immediately
- Root causes that are driving customer dissatisfaction, escalations, chasers
For more information on generating insight and building reports, click here.
8. Implementing automation
The platform enables downstream automation by creating a queue of communications that can be read by a robot.
These queues are driven by the confidence thresholds levels. Setting a threshold means that for the verbatim to enter the queue, the platform must predict that label with a confidence that is equal to or greater than the threshold you set.
For more information on creating and managing streams (formerly known as triggers), click here.
For detailed overview of the UiPath<>Communications Mining automation framework, click here.
9. Additional resources to help you get started
- Introduction to Communications Mining here
- How to navigate the platform here
- Key concepts and terminology here
- Quick training how-to’s here
- FAQs, Tips & Tricks here
Previous: How and where Communications Mining can be deployed | Next: Technical support