v1.1.0
Features
Semantic Clustering
We've introduced a new class that provides the ability to perform Semantic Clustering on Training Phrases that are currently in Dialogflow CX, or that you plan to put in Dialogflow CX. The motivation behind this class is to provide a base "engine" for building tools that allow Bot Builders and Designers with easier methods for determining where their Intents may be "heavy" or "light" with training data. By performing Semantic Clustering, you can quickly pinpoint clusters of Training Data to be groomed from your CX Agent that could be causing conflicts with other Intents.
We'll be extending this class with more tools in the future.
Try out the new features here: SemanticClustering
Utterance Generator
We're excited to introduce another new class that enables users to quickly generate synthetic Training Phrases based on a single input phrase. The motivation behind this class was to provide a way for Bot Builders and Designers to "bootstrap" their Dialogflow CX Agents with an appropriate amount of training data to move to a Production deployment. Once in Production, synthetic data could be swapped out with "real-world" data as it is collected and analyzed for a more fine tuned experience.
In addition, this class can be used to "bolster" Intents that have very few Training Phrases. This is especially helpful when you have high variance in your Training Phrase distribution among Intents, in particular when the # of Training Phrases in your smallest Intents is not at least 10% of the # of Training Phrases in your largest intents.
We'll be extending this class with more tools in the future.
Try out the new features here: UtteranceGenerator
Test Cases (Sync and Async)
Test Cases are finally here! SCRAPI now has support for all Dialogflow CX Test Case features and functionality in both synchronous and asynchronous classes. The motivation behind including the Async class was primarily due to the large scale at which Test Cases can grow to, which can cause running a large set of Test Cases to take quite some time. Allowing these to be run Asynchronously can provide the developer more flexibility to retrieve the results when they are ready, rather than waiting for a large batch of them to return all at once.
You can find the new classes here:
TestCases
TestCasesAsync
Bulk Update Entity From DataFrame
Added a ability to provide bulk updates to Entities via DataFrame, similar to the existing bulk create functionality
Bug Fixes
- Fixed a bug in
core/conversations.py
that forced the input ofcreds_path
arg
Enhancements
- Added
language_code
support toTransitionRouteGroups.update_transition_route_group()
- Added
language_code
support toEntityTypes.create_entity_type()
- Added
language_code
support toDataframeFunctions.bulk_create_entity_from_dataframe()
- Added
language_code
support toEntityTypes.update_entity_type()
Docs
- Updated docs for
EntityType.create_entity_type()
- Updated docs for
DataframeFunctions.bulk_create_entity_from_dataframe()
- Updated
requirements.txt
Misc.
- README updates
- Minor .gitignore updates