Scores, or their derivative, encompass a massive amount of data that can be used to categorize and filter products. We use the entire spectrum and distribution of scores to accurately rank objects according to certain critieria. We developed proprietary algorithms to aggregate scores of different types and from different sources.
A system to make customized recommendations has to be smart in accurately predicting your preferred taste, as well as visionary enough to recommend you products, movies, books, you don’t know yet, but will love.
We use a hybrid recommendation system to achieve just that. Our innovative algorithms (under development) are aimed at personalizing scores depending on your taste and social network.
The Global Wine Score has been the first project developed by our team. It is a worldwide rating which assesses wines with a single score, providing comprehensive and comparable information for all industry players.
It is an average adjusted score aggregating the major wine critics. It takes into account their ways of rating and their respective scales to provide an indicator minimizing the experts subjectivity.
A website has been developed globalwinescore.com and it is already in production used daily by users worldwide.
Scorelab collaborates with IECB (Institut Européen de Chimie et Biologie) on a genomic study. It is a large meta-analysis of multiple genomics experiments on the nematod Caenorhabditis elegans. This analysis aggregates 1600 runs of experiments using RNA sequencing technologies. The data used comes from published studies sharing their datasets on the NCBI (National Center for Biotechnology).
The goal of this work is to understand the functions of genes and the cells’ types in which genes are active. Our statistical tools allow the comparison of experiments with different degrees of sensibility. This work aims to reveal unexpected correlations between genes or experiments. This is still under development.
Scorelab develop algorithms for personalized recommandation of wines. We aggregate our experiences in wine data collection and analysis as well as our skills in machine learning and recommander systems. It allows us to develop and train machine learning models specifically for wine tastes. We learn individuals taste and wines characteristics to generate a personalized wine score.
This score reflects the wines you will probably love and the ones you will have in disgust. It is then applied to your user profile to recommend you wines. A prototype has been developed to test its relevance. Further development are in progress for commercial use.
Scorelab collaborates with ICML 2017 (International Conference on Machine Learning), which develops an application for recommending scientific papers to participants.
We work on the personalized recommender engine of this application. The model in use is an extension of Collaborative Topic Regression.
It is based on both topic modeling of the papers textual content and the users 'likes' on the papers. The application is an open source project and is expected for August 2017.
Do you think you can make a difference ?
See below ↓
If you think you can fit our team and help us grow faster. Please feel free to apply.