Niko Strongioglou

Tasty!

Development Language

Java 1.8

Technology Stack

• Spring/Spring Boot
• Elasticsearch
• Mahout
• Indico.io
• Twilio

Brief Description

Tasty is a recommendation engine designed and developed for the Upstream hackathon competition by out team the Fant4stic100. We wrapped a machine learning system , built on Mahout, around a recipe data set and linked it with an NLP library (indico.io). Various rating categories and NLP data combined together to produce an accurate recommndation for the user. Alongside that implemetation the infrastructure we built provides a means of integrating that recommndation with a consumer service that serves this recipe to your mobile phone either through a google platform or through a simple SMS.

The recommendation engine is based on Pearson correlation similarity and Log likelyhood similarity to produce user to item and item to item recommendations.