This is an example chatbot demonstrating how to build AI assistants for financial services and banking. This starter pack can be used as a base for your own development or as a reference guide for implementing common banking-industry features with Rasa. It includes pre-built intents, actions, and stories for handling conversation flows like checking spending history and transferring money to another account.
Table of Contents
- SCROE Chatbot
It is advisable to create a virtual environment to to avoid version mismatch issues.
Windows 10:
- Create and switch to the project directory:
mkdir scroe-chatbot
cd scroe-chatbot- Create and activate virtual environment named ‘myenv’ using Anaconda prompt:
conda create --name myenv
conda activate myenv- Install required libraries:
#for python version 3.7.x
conda install ujson
conda install tensorflow- Install Rasa:
pip install rasa==2.4.0 - Clone the repo:
git clone ‘bitbucket repo’
cd ‘bitbucket repo’Linux:
- Creating a new directory named venv:
python3 -m venv ./venv- Activating the virtual environment:
source ./venv/bin/activate- Install required libraries:
#for python version 3.7.x
pip install ujson
pip install tensorflow- Install Rasa:
pip install rasa==2.4.0 - Clone the repo:
git clone ‘bitbucket repo’
cd ‘bitbucket repo’Run:
pip install -r requirements.txtTo install development dependencies:
pip install -r requirements-dev.txt
pre-commit install
python -m spacy download en_core_web_md en
python -m spacy link en_core_web_md enFor spaCy
- Download the model:
python3 -m spacy download en_core_web_mdOR
#spaCy link- Link the model
python3 -m spacy link en_core_web_md enFor BERT:
Install the Transformers library using:
pip install transformersDuckling is a pre-trained model which is used to extract amounts of money, numbers, and dates.
It runs as a separate service, in its own container or server.
- Pull
rasa/ducklingdocker image:
docker pull rasa/duckling- Run the
ducklingserver in a new terminal window:
docker run -p 8000:8000 rasa/duckling- We will point to this address instead of localhost in our
config.ymlfile:
- name: DucklingEntityExtractor
url: http://duckling.rasa.com:8000
dimensions:
- amount-of-money
- time
- numberTo run the chatbot.
- Train the model:
rasa train- Start the Rasa open source server on port 5005:
rasa shell -p 5005- Start the action server in a new terminal window:
rasa run actions --port 5056To test the chatbot.
- Run through test conversations to make sure that both NLU and Core make correct predictions:
rasa test- Split data into a test set or estimate how well your model generalizes using cross-validation:
rasa test nlu -u data/nlu.md --config config.yml --cross-validation- Comparing multiple NLU Pipeline Configurations:
rasa test nlu --nlu data/nlu.yml --config config_1.yml config_2.yml config3.ymlThe language models that were considered for our chatbot are:
- Spacy NLP
- BERT
- Roberta
The 4 NLU Pipeline configurations considered are:
config.yml:
version: "2.0"
language: en
pipeline:
- name: WhitespaceTokenizer
- name: RegexFeaturizer
- name: LexicalSyntacticFeaturizer
- name: CountVectorsFeaturizer
- name: CountVectorsFeaturizer
analyzer: "char_wb"
min_ngram: 1
max_ngram: 4
- name: DIETClassifier
epochs: 100
- name: FallbackClassifier
threshold: 0.7
- name: DucklingEntityExtractor
url: http://duckling.rasa.com:8000
dimensions:
- amount-of-money
- time
- number
- name: SpacyNLP
model: "en_core_web_md"
case_sensitive: false
- name: "SpacyEntityExtractor"
# Note: It is not possible to use the SpacyTokenizer + SpacyFeaturizer in
# combination with the WhitespaceTokenizer, and as a result the
# PERSON extraction by Spacy is not very robust.
# Because of this, the nlu training data is annotated as well, and the
# DIETClassifier will also extract PERSON entities.
dimensions: ["PERSON"]
- name: EntitySynonymMapper
policies:
- name: AugmentedMemoizationPolicy
- name: TEDPolicy
epochs: 40
- name: RulePolicy
core_fallback_threshold: 0.4
core_fallback_action_name: "action_default_fallback"
enable_fallback_prediction: Trueconfig-spacy.yml:
version: "2.0"
language: en
pipeline:
- name: SpacyNLP
- name: SpacyTokenizer
- name: SpacyFeaturizer
- name: RegexFeaturizer
- name: LexicalSyntacticFeaturizer
- name: CountVectorsFeaturizer
- name: CountVectorsFeaturizer
analyzer: "char_wb"
min_ngram: 1
max_ngram: 4
- name: DIETClassifier
epochs: 100
- name: EntitySynonymMapper
- name: ResponseSelector
epochs: 100
policies:
- name: AugmentedMemoizationPolicy
- name: TEDPolicy
epochs: 40
- name: RulePolicy
core_fallback_threshold: 0.4
core_fallback_action_name: "action_default_fallback"
enable_fallback_prediction: Trueconfig-bert.yml:
version: "2.0"
language: en
pipeline:
- name: WhitespaceTokenizer
- name: LanguageModelFeaturizer
model_name: "bert"
model_weights: "rasa/LaBSE"
- name: DIETClassifier
epochs: 100
- name: FallbackClassifier
threshold: 0.7
- name: DucklingEntityExtractor
url: http://duckling.rasa.com:8000
dimensions:
- amount-of-money
- time
- number
- name: EntitySynonymMapper
- name: ResponseSelector
epochs: 100
policies:
- name: AugmentedMemoizationPolicy
- name: TEDPolicy
epochs: 40
- name: RulePolicy
core_fallback_threshold: 0.4
core_fallback_action_name: "action_default_fallback"
enable_fallback_prediction: Trueconfig-roberta.yml:
version: "2.0"
language: en
pipeline:
- name: WhitespaceTokenizer
- name: LanguageModelFeaturizer
model_name: "RoBERTa"
model_weights: "roberta-base"
- name: DIETClassifier
epochs: 100
- name: FallbackClassifier
threshold: 0.7
- name: DucklingEntityExtractor
url: http://duckling.rasa.com:8000
dimensions:
- amount-of-money
- time
- number
- name: EntitySynonymMapper
- name: ResponseSelector
epochs: 100
policies:
- name: AugmentedMemoizationPolicy
- name: TEDPolicy
epochs: 40
- name: RulePolicy
core_fallback_threshold: 0.4
core_fallback_action_name: "action_default_fallback"
enable_fallback_prediction: Truedata/nlu/nlu.yml - contains NLU training data
data/nlu/rules.yml - contains rules training data
data/stories/stories*.yml - contains stories training data
actions.py - contains custom action/api code
domain.yml - the domain file, including bot response templates
config.yml - training configurations for the NLU pipeline and policy ensemble
tests/ - end-to-end tests
Rasa x installation -VDI(hrishikesh) - tar file - pip install Ujson wheel file - pip install
Rasa run actions Rasa x
REST Channel: This UI can be used in any website to run Rasa chatbot. Add the index.html file, css and js folders in the rasa project folder and make the mentioned changes.
Steps:
- Since this Chat UI communicates with the Rasa server using
restchannel, make sure you have addedrestchannel in thecredentials.ymlfile - Once you have developed your bot and you are ready to integrate the bot with the UI, you can start the Rasa server using the below command:
rasa run -m models --enable-api --cors "*"- Add the Website Name or Website IP (localhost) in "js/script.js" file:
http://localhost:5005/webhooks/rest/webhookSOCKET.IO: Steps
- Add the below configuration in the credentials.yml file:
socketio:
user_message_evt: user_uttered
bot_message_evt: bot_uttered
session_persistence: true- Start the bot using the below command:
rasa run -m models --enable-api --cors "*" --debug- Create a webpage and add the code from the chat widget in the body tag of your web page.
- Now change the socket url port number to our rasa server's port number:
url: http://localhost:5005/A Spring Boot example app that shows how to implement single sign-on (SSO) with Spring Security's SAML DSL and Okta.
Please read Get Started with Spring Boot, SAML, and Okta to see how this app was created.
Prerequisites: Java 8.
Okta has Authentication and User Management APIs that reduce development time with instant-on, scalable user infrastructure. Okta's intuitive API and expert support make it easy for developers to authenticate, manage and secure users and roles in any application.
To install this example application, run the following commands:
git clone https://github.com/oktadeveloper/okta-spring-boot-saml-example.git
cd okta-spring-boot-saml-exampleThis will get a copy of the project installed locally. To install all of its dependencies and start the app, run:
./mvnw spring-boot:runThe first thing you'll need to do is create an Okta account at okta.com/integrate/signup. After activating your account, log in to it. If you just created an account, you'll see a screen similar to the one below.
Click on < > Developer Console in the top-left corner and switch to the Classic UI. If you see a screen like the following, you're good to go! The reason you need to use the Classic UI for this tutorial is because we haven't yet added SAML support to the Developer Console.
Click Add Applications in the top right to continue. This will bring you to a screen with a Create New App green button on the left.
Click the button and choose Web for the platform and SAML 2.0 for the sign on method.
Click the Create button. The next screen will prompt you for an application name. I used "Spring SAML”, but any name will work.
Click the Next button. This brings you to the second step, configuring SAML. Enter the following values:
- Single sign on URL:
https://localhost:8443/saml/SSO - Audience URI:
https://localhost:8443/saml/metadata
Scroll to the bottom of the form and click Next. This will bring you to the third step, feedback. Choose "I'm an Okta customer adding an internal app” and optionally select the App type.
Click the Finish button to continue. This will bring you to the application's "Sign On” tab which has a section with a link to your applications metadata in a yellow box. Copy the Identity Provider metadata link as you'll need it to configure your Spring Boot application.
The final setup step you'll need is to assign people to the application. Click on the Assignments tab and the Assign > Assign to People button. You'll see a list of people with your account in it.
Click the Assign button, accept the default username (your email), and click the Done button.
Update src/main/resources/application.properties, to have the metadata-url for the app you just created. Make sure to use the "Identity Provider metadata" value you copied earlier (hint: you can find it again under the "Sign On” tab in your Okta application).
server.port = 8443
server.ssl.enabled = true
server.ssl.key-alias = spring
server.ssl.key-store = classpath:saml/keystore.jks
server.ssl.key-store-password = secret
security.saml2.metadata-url = <your metadata url>Start the app using your IDE or mvn spring-boot:run and navigate to https://localhost:8443. If you're using Chrome, you'll likely see a privacy error.
Click the "ADVANCED" link at the bottom. Then click the "proceed to localhost (unsafe)” link.
Next, you'll be redirected to Okta to sign in and redirected back to your app. If you're already logged in, you won't see anything from Okta. If you sign out from Okta, you'll see a login screen such as the one below.
After you've logged in, you should see a screen like the one below.
This example uses Spring Security SAML DSL to integrate with Okta. Kudos to the project's spring-security-saml-dsl-sample for its instructions.
Please post any questions as comments on the blog post, or visit our Okta Developer Forums. You can also email developers@okta.com if would like to create a support ticket.
Apache 2.0, see LICENSE.