SwingR Open Source Cloud Recommender System

SwingR is a Machine Learning Cloud Based Recommender System API for real time E-commerce solutions. SwingR utilizes Neural Networks to predict what kind of products the user is looking to purchase. Based on his/her clicking habits we will train our model to predict what to recommend to the user.

Goals For this Project

• Develop a user freindly API list of endpoints which e-commerce platforms can easily implement into their systems
• Experiment with multiple neural network architectures for recommendation, the e-commerce platform should have the ability to choose the kind of neural network they would like to use
• Perform multiple data transformations and store procedures, to ensure a better recommendation each time a user utilizes the eccomerce website
• Deploy our model to Google Colab is
• Implement a REST API to transact data between e-commerce vendor and our platform
• Implement a Javascript cdn hotlink for web developers to utilize in their HTML scripts
• Implement multiple factor suggestions Descision Trees + Neural Networks

Prerequizits

• Knowllege about Tensorflow
• Python Basics
• REST API implementation
• JSON
• Neural Networks
• Google Cloud Services
• Mathematics (Calculus and Lineer Algebra)

Getting Started (Development)

There are multiple dependencies to install, please install the follwoing and then run python start.py

• Tensorflow
• Numpy
• Flask
• sqllite3
• csv
• MongoDB

API endpoints

The following are some API endpoints that we are working to implement in near duture some have been iplmented already:

http://yourdomain.com/feed/ - This endpoint should be sent a POST request containing the following JSON:

{
    "user_id":"34sd",
    "user_sessison":"123w",
    "user_age":27,
    "date_submitted":"14 September 2018",
    "time_submitted":"8:30pm",
    "purchase_income_todate":"$234",
    "search_activities":[{
        "search_id":"23",
        "search_data":"toys for kids",
        "time_stamp":"7:50pm",
        "results_link":"https://www.lazada.co.th/catalog/?q=kids+toys&_keyori=ss&from=input&spm=a2o4m.home.search.go.1125719cRd99FD"
    },
    {
        "search_id":"25",
        "search_data":"Remote control car",
        "time_stamp":"8:11pm",
        "results_link":"https://www.lazada.co.th/catalog/?q=remote+control+car&_keyori=ss&from=input&spm=a2o4m.searchlist.search.go.27cf5478bpLNl6"
    },
    {
        "search_id":"29",
        "search_data":"pots and pans",
        "time_stamp":"9:30pm",
        "results_link":"https://www.lazada.co.th/catalog/?q=pots+and+pans&_keyori=ss&from=input&spm=a2o4m.searchlist.search.go.186f2ca0fKEpwi"
    }
]
}

http://yourdomain.com/suggest/{user_id} - This endpoint should be sent a GET request with the user id such that the system can scan for available sugegstions for that user, before a suggestion is ready a user must have had at least 100 search entries registered in the system, after which this endpoint can be called after a specified number of minutes

Authors

This project what initially started by myself and @aakashsplinter, we have created this project with the intention of creating a community based open source project which we all can commit to and benifit from using and exandiing your skillset.

Contributing

Please feel free to create issues and push code to this respository as and when you like, we always like adding new ideas to this respository to ensure a mix of ideas and functionality. Please ensure you make detailed commits so that we have an idea of what you where trying to achieve, before pushing to production Branch.

Versioning

Please utilize the versioning methodology found in this respositiory https://semver.org/