LLM_TRANSLATION-CHATBOT

Description

This project focuses on integrating a large language model (LLM) to accomplish two main tasks: language translation and chatbot development from website data.

• Part 1: Language Translation

  The system uses the Gemini API to translate text into Vietnamese. It can process either a single text or a list of multiple texts while ensuring that if the text is already in the target language, it remains unchanged.

• Part 2: Web-Based Chatbot

  The chatbot requires collecting information from a specific website, preprocessing the data, indexing it for search, and developing a chatbot capable of answering questions based on the gathered content. It utilizes NLP to find the most relevant text from the website data and then feeds it into the LLM to generate accurate responses.

  Here is the link to the website: presight

This project integrates web crawling, NLP (natural language processing), and AI (artificial intelligence) to build an intelligent chatbot system and facilitate automatic translation.

Results

1. Translation

• Supports translating both single texts and lists of texts into Vietnamese.

• Detects the input language and keeps the text unchanged if it is already in Vietnamese.

  Example

  • Input: "Hello" → Output: "Xin chào"

  • Input: ["Hello", "How are you?", "Tôi là sinh viên"] → Output: ["Xin chào", "Bạn khỏe không?", "Tôi là sinh viên"]

  

2. Chatbot

• Collects data from Presight, processes it, and creates a search index.

• Uses NLP to find the most relevant passage from the website data and generates answers using LLM.

  Example

  • Question: "What is Presight’s privacy policy?"

  • Chatbot response: "The policy is a Privacy Policy that explains how Presight collects, uses, and discloses information about their customers and website visitors."

  

3. Web Interface for Chatbot

• Objective: Enable users to easily query information from Presight's policies through an online chatbot.

• KEY FEATURES

  • Simple and intuitive interface: users can enter questions in the chat box and receive instant responses.

  • Direct connection to chatbot backend: the system sends queries to the backend, processes them using NLP, and returns relevant answers.

  • Displayed information with source citations: responses may include citations from Presight’s policies, allowing users to verify the information.

  • Conversation history: users can review previous questions to track information.

• Demo & Illustrations: The demo video is attached to the project.

  

Contributors

We sincerely thank all team members for their dedication and time in developing this project. Your collaboration and sense of responsibility have contributed to achieving the best possible results.

Name	Major	University
Phan Binh Phuong	Data Science	University of Science (VNUHCM)
Huynh Thao Quynh	Data Science	University of Science (VNUHCM)
Nguyen Ngoc Thanh Thu	Data Science	University of Science (VNUHCM)
Ho Tran Anh Thu	Data Science	University of Science (VNUHCM)