🤖 NLU Chatbot Project (Work in Progress)

This is a Natural Language Understanding (NLU) project written in Python, where I aim to build a chatbot that can intelligently respond to customer inquiries based on their utterances.

Note: This project is still in development! I’m currently exploring two different NLP libraries: spaCy and Stanza. The code isn’t finalized yet, and I’m experimenting and learning as I go.

---

Project Goal

The goal of this chatbot is to understand customer utterances from a dataset and provide appropriate responses. The bot doesn’t use predefined intents — instead, it learns patterns and clusters them based on semantic similarity.

---

What I'm Doing (Step-by-Step)

1. Preprocessing customer utterances from a dataset
2. Tokenizing the text using both spaCy and Stanza
3. Creating custom stopword lists
4. Mapping certain words (e.g., synonyms or brand-specific terms) to standard forms
5. Lemmatizing the tokens for better generalization
6. Vectorizing utterances using Sentence-BERT (SBERT)
7. Using KMeans Clustering to group similar utterances
8. Assigning appropriate chatbot responses based on clusters

---

Approaches

I'm testing two NLP pipelines:

• chatbot_spacy.py – Based on spaCy
• chatbot_stanza.py – Based on Stanza

Each script is a work-in-progress and may evolve over time.

---