Logo Lanfrica
  • Accueil
  • Atlas
  • Analyses
  • Documentation
  • Sign in

© 2026 Lanfrica. Tous droits réservés. Tous les droits d'auteur des ressources affichées sur le site Web Lanfrica appartiennent aux détenteurs de droits d'auteur d'origine, sauf indication contraire explicite.

ahmedkhairysaleh-boop/etisalat-rag-chatbot

Domaine:

natural language processing

Type de record:

software
Créateur:
ahm
Hôte:
Bilingual RAG chatbot over e& Egypt package documentation, built with LangGraph and ChromaDB # e& Egypt Knowledge-Base Chatbot A bilingual (Arabic/English) retrieval-augmented chatbot that answers questions about e& Egypt mobile packages, internet plans, prices and service codes, grounded entirely in the company's official documentation. Built for the **e& Summer Internship Program** by Ahmed Khairy. --- ## What it does Ask a question in Arabic, English, or a mix of both. The system retrieves the relevant passages from the knowledge base and answers from them, citing which source each figure came from. If the documentation does not cover the question, it says so rather than guessing. ``` > How much does Emerald 430 cost per month? The Emerald 430 package costs 430 EGP per month before taxes [1]. > عايز اعرف سعر باقة حكاية 46 سعر باقة حكاية انترنت 46 هو 46 جنيه [1]. (ميني حكاية انترنت 46 بصلاحية 3 أسابيع سعرها 46 جنيه أيضاً [2].) > ايه تفاصيلها حكاية إنترنت 46 هي باقة شهرية، تفاصيلها كالتالي: · السعر: 46 جنيه · الإنترنت: 1,250 ميجابايت · وحدات المكالمات والرسائل: 300 وحدة ... ``` --- ## Architecture ```mermaid graph TD A[Customer question] --> B[understand] B -->|greeting or thanks| C[smalltalk] B -->|real question| D[retrieve] D -->|relevant context| E[generate] D -->|nothing usable| F[no_context] C --> G[Answer] E --> G F --> G subgraph inside_understand [inside understand] H[detect language] I[rewrite follow-ups using recent turns] J[split multi-product questions] end subgraph components [ ] K[(ChromaDB 447 chunks)] L[Groq gpt-oss-120b] end D -.- K E -.- L I -.- L J -.- L ``` The agent is a LangGraph state machine with two conditional edges. **understand** detects the question's language and prepares the search. Two optional LLM calls happen here: - *Query rewriting* — a follow-up like "what are its details" carries no product name, so it is rewritten using the last four turns into a question that can stand alone. - *Query decomposition* — a question naming two products is split into one search per product. Both are skipped when they …

Visit

github.com

Tasks

question answering

Similaires

Cliff-Ezra/RAG-chatbot-v2zakariae-boui/onssa-rag-chatbotdjpapzin/malawi-rag-sql-chatbotGilbertoBotas/Mozambican-Road-Code-Chatbot-RAGFERRADJI-TAREK/chatbot-djezzy-rag-finetuningojoluwasegun/Nigeria-Tax-Reform-RAG-Chatbot-HuggingFace

Cliff-Ezra/RAG-chatbot-v2

This repository is for the implementation of a RAG based chatbot for Kenya Law. It is the second ver

zakariae-boui/onssa-rag-chatbot

Customer-assistance RAG chatbot for the ONSSA website (Morocco) — Streamlit, Ollama (Mistral 7B), FA

djpapzin/malawi-rag-sql-chatbot

# Dziwani - Infrastructure Transparency Chatbot A specialized chatbot for querying and exploring Ma

GilbertoBotas/Mozambican-Road-Code-Chatbot-RAG

RAG-powered assistant for the Mozambique Road Code. Ask questions about traffic rules, penalties, an

FERRADJI-TAREK/chatbot-djezzy-rag-finetuning

This project involves designing an intelligent chatbot to improve customer support at Djezzy, a majo

ojoluwasegun/Nigeria-Tax-Reform-RAG-Chatbot-HuggingFace

This project is an intelligent Question-Answering (QA) chatbot built to help users explore and under