IT Career Bot is a project for desmontrating how to build a personalized chatbot based on context-aware knowledge base. The chatbot can help students know how to prepare for a new career by providing dynamic learning paths according to user interest and profile.
The source code is structured as the below:
- api-server: Contains Python scripts to run Flask application for serve user requests.
- chat-widget: A ReactJS applications provide an interface that allows user can text and get response from the chatbot.
- ontology: Contains raw data that is collected from different sources (Ex. Edx, Coursera, ...) and scripts for creating career counseling ontology.
- survey-analysis: Contains raw data that is questions is collected from social networks (Quora, Stack overflow, Stack exchange) to understand what user demand need to in a career counseling service.
- rasa: Contains pipeline configuration and dialog management for intent classification and entity extraction.
The chatbot architecture consists of 3 main components:
- Conversation handler is implemented by using Rasa pipeline for intent classification and entity extraction.
- Response generator is the component that generates response based on user contexts and domain knowledge.
- Knowledge base represents entities related to career counseling services such as occupation and associated skills, courses and users contexts.
- Virtual environment
- We use Anaconda (https://www.anaconda.com/) to create an isolated virtual environment for interpreting and executing Python scripts. It also provides many modules and libraries to enable a flexible approach when programming chatbot functions.
- A Python web framework allows create production in a short time without considering the low level of web programming.
- Firestore communication
- We use Firebase-admin (https://pypi.org/project/firebase-admin/) module to query and retrieve data from Firestore (A No SQL cloudd database is developed by Google)
- Owlready2 (https://owlready2.readthedocs.io/en/v0.34/)
- This Python library provide many functions to create, update and query on an ontology. We recommend use version 0.34 to avoid conflicting in the setup process.
- Step 1. Execute Rasa component by using the below command:
rasa run --enable-api --cors "*" --port 5005 --debug --credentials credentials.yml
- Step 2. Running Rasa action server to serve single actions for each request from dialog management
rasa run actions
- Step 3: Running the Flask server to handle user requests
flask run