Previous knowledge of Rasa Framwework is expected
To run the model created just type in the command line:
cd Scriptsrasa run actionsrasa shell --endpoints.yml
- 2 and 3 must run simultaneously in the different terminal windows
The desing of the chatbot is better represented in the two tables below:
| Element | Details |
|---|---|
| Name: | MPMD pal |
| Type: | Frequently Asked Questions (FAQ) |
| Purpose: | Solve questions based on MPMD HTW Website |
| Role: | Support MPMD staff by answering basic queries to any in-terested person and referring them to contact a given mail if the bot cannot answer the questions. |
| Audience: | Any interested person in the program, commonly people between ~25-35 years. |
| Features: | Text-only interphase. |
| Preferred devices: | Any device capable of using Facebook, commonly desktops, laptops, mobiles, and tablets. |
| Development framework: | Rasa 1.1.4 |
| Programming language: | Python 3.6 |
| Element | Details |
|---|---|
| Conversational AI | |
| Type: | Retrieval-based |
| Intents: | 20 |
| Utterances per intent: | ~100 |
| Stories: | 1 |
| Rasa NLU | |
| Pipeline: | Supervised Embeddings |
| Tokenizer: | Whitespace |
| Intent Classifier Model: | Neural Embedding |
| Rasa Core | |
| Dialogue predictor model: | Mapping Policy |
| Iniciative: | Managed by the user |
| Restart dialogue: | Mapping Policy |
| Grounding: | Fallback policy |
- The Chatbot's architecture is designed to exclude all mapped intents from the dialogue predictions.
- After each action, an "action_listen" is forced.