Third-party Dependencies

The following table lists the third-party software libraries and their versions.


Library	Version
python	3.8.15
pyjwt	2.4.0
pandas	1.3.5
python-dateutil	2.8.2
html5lib	1.1
scikit-learn	1.0.2
lxml	4.7.1
transformers	4.26.1
tensorflow
Keras	2.6.0
numpy	1.22.2
oci	2.70.1
requests	2.28.1
schedule	1.1.0

The base code is developed and passed along with documentation to the customers for solving their problems by integrating into other services using the B2C Service's capabilities under the Universal Permissive License (UPL).

For more information, refer to UPL License. The base code can be customized according to the use case , which can be used to solve the business problems specific to the customers. The customers can use this as it is or as a starting point to solve their problems using this integration.

Note: This code is intended for education or testing purposes for the B2C Service customers.

This accelerator makes use of the "OCI Language Classification Model" and historical chat conversation with the help of the EmoRoBERTa model to identify customer sentiments. Once the negative sentiment predicted by this model reaches the set threshold, the accelerator flags the chat as "Needs Attention" This accelerator also outlines the mechanism of auto-correcting the model in case of incorrect prediction and uses agent feedback as a way forward.

Note: This accelerator uses EmoRoBERTa, a third-party model provided under the terms of an open source license without warranty of any kind. Please refer to the Third Party Notices.txt in the code repository for details. By proceeding to configure this accelerator, you confirm that you understand and agree to the terms and the risks associated with the third-party model.

Live Chat Sentiment Analysis Accelerator makes use of Custom Process Model (CPM) and extensions feature of the B2C Service application to classify sentiments for chat conversation with the help of OCI custom language classification model.