-
Description: An intelligent web interaction tool that harnesses the power of RAG (Retriever, Aggregator, Generator) flow to facilitate seamless communication with websites.
-
Functionality:
- Upon receiving a webpage link, WebChatter swiftly scrapes the webpage, extracts its textual content, and divides it into manageable chunks.
- These chunks are then embedded with metadata and stored in a Pinecone vector database.
-
User Interaction:
- Once the index is established, users can engage in conversational queries.
- By inputting a query, WebChatter generates embeddings for the query and calculates cosine similarity scores between these embeddings and the embeddings of the webpage chunks.
- The chunks exhibiting the highest similarity scores are selected and passed to the Language Model (LLM) along with the query, facilitating accurate and targeted responses.
-
**UI Demo
-
- Above is the image where User have to input the Webpage link.
-
- Above is the image where User will input his query and chat with the Webpage
-
- Above is the image where User receives his response from the bot.