ghostai1 commited on
Commit
d45df4b
·
verified ·
1 Parent(s): eda95b4

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +40 -13
README.md CHANGED
@@ -1,13 +1,40 @@
1
- ---
2
- title: CXRAG
3
- emoji: 🏃
4
- colorFrom: green
5
- colorTo: indigo
6
- sdk: gradio
7
- sdk_version: 5.28.0
8
- app_file: app.py
9
- pinned: false
10
- short_description: Junk DATA RAG CAG
11
- ---
12
-
13
- Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ CX Bot Demo
2
+ A RAG-based customer experience (CX) bot deployed on Hugging Face Spaces (free tier). Demonstrates junk data cleanup and client data validation for high-quality, multilingual CX solutions in SaaS, HealthTech, FinTech, and eCommerce.
3
+ Features
4
+
5
+ RAG Pipeline: Retrieves FAQs using all-MiniLM-L6-v2 and FAISS, delivering accurate responses.
6
+ Data Cleanup: Removes nulls, duplicates, and low-quality FAQs (e.g., short answers) to ensure reliable outputs.
7
+ Performance Plot: Visualizes latency and accuracy with Matplotlib/Seaborn to monitor data quality.
8
+ Gradio UI: User-friendly interface for querying, viewing FAQs, and checking cleanup stats.
9
+
10
+ Setup
11
+
12
+ Clone this repo to a Hugging Face Space (free tier, public).
13
+ Create requirements.txt with listed dependencies.
14
+ Upload app.py (includes embedded sample FAQs).
15
+ Set Space to run with Python 3.9+ and no GPU.
16
+
17
+ Usage
18
+
19
+ Enter a query (e.g., “How do I reset my password?”) in the Gradio UI.
20
+ View the bot’s response, retrieved FAQs, cleanup stats, and RAG pipeline plot.
21
+ Example output:
22
+ Response: “Go to the login page, click ‘Forgot Password,’...”
23
+ Cleanup Stats: “Cleaned FAQs: 3 (removed 2 junk entries)”
24
+
25
+
26
+
27
+ Data Cleanup
28
+
29
+ Preprocess FAQs: Removes nulls, duplicates, and answers <20 characters to ensure high-quality data.
30
+ Query Validation: Rejects empty or short queries (<5 characters) for reliable input.
31
+ Why It Matters: Clean data is critical for accurate, scalable CX solutions, ensuring robust performance for enterprise Partners.
32
+
33
+ Technical Details
34
+
35
+ Stack: Python, Hugging Face (all-MiniLM-L6-v2), FAISS (CPU), Gradio, Pandas, Matplotlib, Seaborn.
36
+ Free Tier: Lightweight design (no GPU, small model) for Hugging Face Spaces.
37
+ Extensibility: Adaptable for CRM integrations (e.g., Salesforce) and cloud deployment (e.g., AWS Lambda).
38
+
39
+ Purpose
40
+ Built to demonstrate expertise in designing, building, and deploying CX bots with a focus on data quality, suitable for AI-driven customer experience platforms.