Data Library

Today, data scientists have algorithms at their fingertips, so why not data?

One platform to rule them all

Knowing exactly what data is available is a relentless challenge. Using a retrieval engine, we collect data and metadata every day to feed our platform. We can screen any information source, from open data platforms to data brokers, satellite and aerial imagery producers and web pages.

Highly qualified metadata

The metadata we get from the platforms isn’t standardized or detailed. That’s why our scripts run through every dataset in our platform to get as much information as possible about the content, right down to the columns. We classify, order and enhance the data using metadata and an efficient search engine.

More than a catalog, a library

A catalog simply redirects you to remote data stored in another server. This data might be slow to download or even unavailable. That is why we download all the data which feeds our platform into a library – to make sure it is available to you anytime and anywhere. Customized data visualization technologies enable us to travel though datasets in any format e.g. tabular data, geolocalized data, images and texts.

Our Data Library in figures

114
platforms

crawled every day

1100.1
thousand

files gathered

15400
Thousand

columns qualified

files collected are geolocalized 0

files collected are geolocalized by nam.R 0

Data Science

Build our own geolocalized database

Natural Language Processing

To extract structured data from the text files in our Data Library, we use Natural Language Processing algorithms. We’ve developed specific algorithms for information retrieval and named entity recognition that identify which texts elements correspond to which names in which location. These algorithms were trained using a unique corpus built by nam.R.

Computer Vision

To extract information from satellite, aerial or terrestrial images, we have developed expertise in computer vision. Our technology is based on deep learning.
We have also built a very effective image annotation method to produce our own learning bases.
These techniques allow us to detect elements like cars or swimming pools on aerial images or to count windows on photos of facades.

Machine Learning

To be more efficient and optimize computational costs, Nam.R has developed new approaches to machine learning that are adapted to localized data.

Data Visualization

See to Believe

3D Map

Seeing your data is a powerful way to generate meaning and drive concrete actions.

Developed in WebGL based on the three.js  framework, our 3D Map integrates all our data. This 3D GIS shows the data we build every day, in real time.