Improve Entity Extraction for Exercises, Food Types, and Key User Information

Hello, @Akki-jaiswal, I was exploring you project and found that i can contribute in following way by improving the chatbot.

The chatbot could be improved to **better understand user queries** related to exercises, food intake, and other fitness-related information.
This issue proposes developing robust entity extraction capabilities that can accurately identify and normalize such data.

Here’s a plan we could follow to improve this: {{ Ideas to Explore }}

- _Rule-Based Pattern Matching_ – Start simple with dictionaries of exercises (push-ups, squats, running) and foods (chicken, pasta, broccoli) + regex/string matching for quick wins.
- _NER Models_ – Use spaCy or Hugging Face to detect entities, and train custom ones for EXERCISE and FOOD so it understands domain-specific terms.
- _ML Classifiers_ – If we want more accuracy, train CRF/neural models on labeled data to learn where entities start and end.
- _Keyword + Context Analysis_ – Look for trigger words like “ate,” “did,” or “performed” to better guess what’s coming next.
- _Training Data_ – Collect sample user queries, label them carefully, and cover different ways people talk (e.g., “went for a jog,” “jogging,” “ran”).
- _Entity Normalization_ – Make sure “pushups,” “push ups,” and “push-ups” all mean the same thing.

Would love to discuss this approach with you.
If everything sounds good, should I go ahead and start working on a first version of this?