Building RAG Chatbot for Movie Recommendations with Qdrant and Open AI

For the Qdrant and Open AI platform, this workflow automates movie recommendations by integrating data from GitHub and processing it with AI. It efficiently extracts movie details, generates embeddings, and queries a vector database to provide personalized movie suggestions based on user preferences. This streamlined process enhances user experience by delivering tailored recommendations quickly and accurately.

7/8/2025
27 nodes
Complex
manualcomplexgithubextractfromfilelangchainexecuteworkflowtriggersplitoutaggregatesticky noteadvancedfilesstorageapiintegration
Categories:
Complex WorkflowManual TriggeredTechnical Infrastructure & DevOpsBusiness Process Automation
Integrations:
GitHubExtractFromFileLangChainExecuteWorkflowTriggerSplitOutAggregateSticky Note

Target Audience

Target Audience


- Movie Enthusiasts: Individuals looking for personalized movie recommendations based on their preferences.
- Developers: Those interested in integrating AI and vector databases for recommendation systems.
- Data Scientists: Professionals seeking to utilize machine learning for enhancing user experiences in media consumption.
- Content Creators: People who want to build engaging applications that provide movie suggestions to their audience.

Problem Solved

Problem Solved


This workflow addresses the challenge of finding relevant movie recommendations based on user preferences. By leveraging a vector database and AI embeddings, it effectively matches user queries with a curated dataset of movies, ensuring that users receive tailored suggestions that align with their tastes and interests.

Workflow Steps

Workflow Steps


1. Manual Trigger: The workflow begins when a user clicks the ‘Test workflow’ button.
2. GitHub Integration: It retrieves a CSV file containing the Top 1000 IMDB movies from a specified GitHub repository.
3. Extract from File: The data is extracted from the CSV to prepare for processing.
4. Embeddings Creation: Using OpenAI's embeddings model, the workflow generates vector representations of movie descriptions.
5. Data Loading: The extracted movie data is loaded into a Qdrant vector store for efficient querying.
6. Chat Trigger: The workflow listens for incoming chat messages to initiate the recommendation process.
7. AI Agent: Upon receiving a message, the AI agent processes the user’s request to understand their preferences.
8. Embedding Requests: It sends requests to OpenAI to create embeddings for both positive and negative examples based on user input.
9. Qdrant Recommendation API: The workflow queries the Qdrant API to retrieve movie recommendations based on the generated embeddings.
10. Meta Data Retrieval: It fetches detailed information about the recommended movies from the Qdrant database.
11. Data Aggregation: The relevant details are aggregated and prepared for presentation to the user.
12. Output: Finally, the top-3 movie recommendations are presented to the user without disclosing the recommendation scores.

Customization Guide

Customization Guide


- Modifying Movie Dataset: Users can change the GitHub repository and file path to point to a different dataset containing movies.
- Adjusting AI Models: Users can switch the OpenAI model used for embeddings or chat responses to suit different performance or budget needs.
- Changing Recommendation Criteria: The parameters for the Qdrant API can be adjusted to refine how recommendations are generated, such as altering the strategy for vector averaging.
- User Interaction: The chat trigger can be adapted to integrate with different messaging platforms or user interfaces to enhance interactivity.
- Memory Management: Users can customize the Window Buffer Memory settings to alter how much context is retained during conversations.