After China and the US, India had the third-largest online shopper base of 150 million in FY21 and is expected to be 350 million by FY26. Keeping in mind the growing demands of modern India, e-commerce is set to change forever. The e-commerce business is totally saturated with everything at our fingertips but that also means it is prone to mishaps. Hence, we make use of what the 21st century is famous for- TECHNOLOGY! to truly transform businesses and how they operate in a fast-paced country like India.
Our main goal is to help businesses make more money and customers gain an enjoyable and personalized shopping experience.
We aim to truly revolutionize the e-commerce shopping experience for everyone, all around our country, in every language possible! We want to give online shoppers the same kind of personalized experience as they would in a traditional in-person interaction. After covid the whole world is online and e-commerce truly helps bring the world at your fingertips hence we make use of GRAPH ML. A highly advanced way to bring the users the best and most relevant things to them.
To help scale businesses to their full potential and reach the right customers and provide them with the most personalized and relevant experience building loyal customer relationships :)
We leveraged the power of graph ML and implemented a robust backend architecture that is able to run on regular mobile apps making it feasible to implement into businesses. We built a backend using janusgraph, gremlin and python and converted it into an API to be able to integrate and push recommendations on our app.
- Machine Learning - Python files
- React Native App
- A reccomendation API service
- Gephi Files + Report
Our approach stands out by leveraging key aspects:
Graph Data Representation: We excel at modeling intricate relationships and connections, enabling a comprehensive understanding of complex data.
Graph Query Insights: We extract valuable patterns and clusters from the data, providing meaningful and actionable insights.
Graph Machine Learning: By utilizing Graph Neural Networks, we achieve precise and accurate predictions, enhancing decision-making capabilities.
Optimal Architectures: Our system ensures efficient processing on commodity-based hardware, optimizing performance while keeping costs reasonable.
Integration with Mobile App/Web: We offer a user-friendly interface, enabling easy access to the system's capabilities through mobile and web applications.
The unique combination of these features empowers our solution to deliver next-generation business capabilities based on the SNAP product co-purchasing networks dataset.
- React Native
- JanusGraph
- Gremlin
- Python
- Gephi
- Visual Studio
- Expo
- RESTAPI
- Figma
- Docker
- Dataset used - SNAP amazon co- purchasing (https://snap.stanford.edu/data/amazon0601.html)
- Installation and implementation of Janus Graph
- Integration of ML into app
- Handling very heavy datasets
- A fully functioning reccomendation ML model based on graph ML dataset
- An end- to - end integrated full stack app
- An ML based reccomendation system
- Personalization of reccomendations
-
Additional Data Sources: Incorporate diverse data sources (e.g., customer demographics, social media, product reviews) for a comprehensive understanding of user preferences, improving recommendation accuracy and relevance.
-
Contextual Recommendations: Consider location, time, and device type to provide context-aware recommendations, enhancing the user experience with situation-specific suggestions.
-
Hybrid Recommender Systems: Integrate collaborative filtering, content-based filtering, and graph-based techniques to offer a diverse range of recommendations and overcome limitations of individual approaches.
-
Explainable Recommendations: Provide transparency and user trust by offering explanations for recommendations, leveraging interpretable graph embeddings or rule-based explanations.
-
Continuous Learning and Adaptation: Implement mechanisms for continuous learning and adaptive recommendations based on user feedback and changing preferences, utilizing online learning and reinforcement learning techniques.
-
Integration with Advanced Analytics: Incorporate advanced analytics capabilities like sentiment analysis, customer segmentation, and lifetime value prediction to gain deeper insights into customer behavior and enable targeted recommendations.
-
Privacy-Preserving Recommendations: Address privacy concerns with techniques such as federated learning and differential privacy to protect sensitive user data while ensuring accurate recommendations.
By considering these enhancements, the graph-based recommendation system for co-purchasing networks can provide more accurate, personalized, and valuable recommendations, leading to increased customer engagement and driving business growth.
Only developers.
Name | GitHub Profile |
---|---|
Parth Katiyar | Github |
Bhurva Sharma | GitHub |
Affaan Kidwai | GitHub |