Multi-Topic Language Intelligence Training
- 01 Jul 2024
- 1 Minute to read
- Contributors
- Print
- DarkLight
- PDF
Multi-Topic Language Intelligence Training
- Updated on 01 Jul 2024
- 1 Minute to read
- Contributors
- Print
- DarkLight
- PDF
Article summary
Did you find this summary helpful?
Thank you for your feedback
Ushur can be utilized to classify or tag inbound emails. A tag or a keyword is assigned to the email based on the intent. Sometimes, other workflow management systems use these tags to drive other automation work. In such scenarios, a . CSV file containing all topics and phrases is used to train the LI model, and hence the name.
A sample scenario looks like this.
In this example result of classification (stored in a Ushur variable) is being matched with the metadata table and the corresponding email address is used in an email module to forward the email.
The table below is a description of the modules and their action in the workflow.
Module | Description |
Email Processor Module | This module will process the inbound e-mail which can be an API or a forwarded email. |
LI Module | This module applies the classification to the e-mail body. The result of this classification (topic) is stored in the Ushur emailvariable.  |
Email Data Set | This is an empty notify module that is trained via bulk phrases with all topics and phrases in one module. In this example, there are four categories i.e. Billing, claims, registration, terminationand .  |
Compute Module | This module assigns the topic (one of the 4 topics names in this example) and classification confidence to their corresponding variables. |
Find Department | In this example, we use the fetch module to find the department in the metadata based on the classification result. Your use case might be different but the concept stays the same. |
Forward to SME | This module will then forward the email to the respective subject matter experts (SME) fetched from the metadata table. |
Was this article helpful?