Our productions


Hyphe

Web content indexation and automatized deployment on OpenStack

Hyphe is a web crawler designed for social scientists, and developped by Sciences-Po médialab.

We added the following features:

  • Automatic textual indexation of web corpuses by multiprocess content extraction and indexation inElasticsearch
  • Automatic deployment of Hyphe server on OpenStack compatible hosting services

A Open Source and Open Data project

  • Digital humanities
  • Python
  • ElasticSearch
  • JavaScript
  • DevOps
  • OpenStack
  • Functional tests of indexation processus
  • Hyphe server to be deployed configuration (Hyphe Browser)
  • Cloud server to be deployed specifications (Hyphe Browser)
  • Functional tests of indexation processus

    Cross
  • Hyphe server to be deployed configuration (Hyphe Browser)

    Cross
  • Cloud server to be deployed specifications (Hyphe Browser)

    Cross

Configuration management database

Exploring a CMDB through ego-centered networks

One of the largest French industrial group indexed their whole CMDB in a Neo4j database. They contacted us to develop an interface to explore this dataset.

The web application is composed of a search engine and a unique page for each node, displaying its neighborhood and metadata. To make the search engine efficient (errors tolerance, searching on multiple fields), we indexed the corpus in an ElasticSearch base.

The frontend is developed with Angular, and the API runs on Node - the whole with TypeScript.

A Custom development project

  • Industry
  • Neo4j
  • ElasticSearch
  • Angular
  • Sigma.js
  • Search page, through the different node types
  • Node page, with its ego-centered network, the list of its direct neighbors and its metadata
  • Fullscreen network exploration
  • Search page, through the different node types

    Cross
  • Node page, with its ego-centered network, the list of its direct neighbors and its metadata

    Cross
  • Fullscreen network exploration

    Cross

HOPPE-Droit

Explore a 19th-20th centuries French law educational works collection

The HOPPE-Droit projects aims to create and publish a collection of French law educational works from the 19th-20th centuries. We designed and developed an exploration tool which help studying the evolutions of the French law through education materials from the 19th century.

The dataset is edited by the CUJAS team in a Heurist database. These data are exported through API to be indexed into ElasticSearch. We made sure to keep data complexity by for instance taking care of dates incertainty and levels of precision. A web application finally allows to explore the dataset under different angles: books, authors, editors, co-publication networks, genealogies...

A Custom development project

  • Digital humanities
  • Data visualisation
  • Conception
  • ElasticSearch
  • React
  • Heurist
  • Sigma.js
  • Faceted search on the collection authors
  • Network of authors-editors linked by educational works
  • Editor's genealogy page summing up association, acquisition...
  • Faceted search on the collection authors

    Cross
  • Network of authors-editors linked by educational works

    Cross
  • Editor's genealogy page summing up association, acquisition...

    Cross

RadioPolice

Visual analysis and semantic extraction of themes of a tweets data-set

We were contacted to semantically analyse a corpus of french tweets. We set up a topic extraction pipe, through terms co-occurrences analysis and CHI² token filtering. We also shared some online tool to explore topic communities, in terms co-occurrences network maps.

David Dufresne and the Mediapart french journal wanted to publish the corpus. We helped set up ElasticSearch and Kibana to forge one query per curated topic, and to get aggregated indicators for the final user interface designed and developed by WeDoData, Etamin Studio and Philippe Rivière / Visions carto.

A Data valorization project

  • Data journalism
  • Python
  • Natural Language Processing
  • Data science
  • Visual analysis
  • ElasticSearch
  • Kibana
  • Co-occurence network of terms from the "(il)légitimité" theme
  • "palet" neihgboors in the co-occurrence network of significant terms
  • Building the theme "outrage" as a search query in Kibana
  • Co-occurence network of terms from the "(il)légitimité" theme

    Cross
  • "palet" neihgboors in the co-occurrence network of significant terms

    Cross
  • Building the theme "outrage" as a search query in Kibana

    Cross

Exhibition-test

Data infrastructure specifications of an interactive exhibition

We designed the data infrastructure for an exhibition which observes its visitors: data flows specifications from data capture to video-walls projecting visualisations, going through analysis, archiving and graphic rendering processes.

The exhibition was canceled due to COVID-19 epidemic. We haven't realized those plans yet.

A Consulting and support project

  • Digital humanities
  • Realtime data
  • Data visualisation
  • Conception
  • Architecture
  • Data infrastucture schema extract
  • Physical infrastucture schema extract
  • Data infrastucture schema extract

    Cross
  • Physical infrastucture schema extract

    Cross

Production monitoring dashboard

Custom Kibana plug-ins development

An industrial actor contacted us to help them distribute dashboards within one of their product. After a brief benchmarking, Kibana felt the best solution, despite missing some key features.

We developed a custom plugin with these features (integrating the dashboards inside a custom web page, with custom styles).

A Consulting and support project

  • Industry
  • Kibana
  • ElasticSearch
  • Dashboard

TOFLIT18

Toflit18 is an online exploratory analysis tool on 18th century french trade by products. We updated this tool created by Sciences Po médialab by optimizing the Neo4j queries and by adding permalinks and a data table which lists and filters trade flows.

A Open Source and Open Data project

  • Digital humanities
  • Neo4j
  • JavaScript
  • React
  • Nantes exports trade flows from 1720 to 1780
  • Classifications coverage ratio pptimization
  • Permalink of 18th century Nantes exports terms networks
  • Nantes exports trade flows from 1720 to 1780

    Cross
  • Classifications coverage ratio pptimization

    Cross
  • Permalink of 18th century Nantes exports terms networks

    Cross

RICardo

RICardo is a research project about international trades in the 19-20th centuries.

We improved the existing web application:

  • Refactoring of the existing visualizations
  • New visualizations of exchange rates and political statuses
  • Permalinks with visualization parameters, on all pages

Read our blog post "Some new visualizations for the RICardo project" to learn more about this contract.

A Data valorization project

  • Digital humanities
  • Data visualisation
  • Conception
  • JavaScript
  • AngularJS
  • This additionnal timeline foster taking the political context into account when analysing historical trade
  • We created a heatmap to compare relative importance of trade partners
  • Exchange rates exploration through small-multiples
  • This additionnal timeline foster taking the political context into account when analysing historical trade

    Cross
  • We created a heatmap to compare relative importance of trade partners

    Cross
  • Exchange rates exploration through small-multiples

    Cross

E-commerce and online payment

We helped developing the payment process of one of the biggest French e-commerce websites, using Clojure and ClojureScript.

A Consulting and support project

  • Industry
  • Clojure
  • ClojureScript
  • Web performance

Contractor for Neo4j

We work on behalf of Neo4j to assist their customers on their graphs projects. We do Neo4j consulting, from data modeling, loading and visualization, to prototypes and full web projects based on modern web technologies.

A Consulting and support project

  • Neo4j
  • Data visualisation
  • Conception

Bibliograph

Bibliograph is an online tool which we created with and for Tommaso Venturini to equip his research on scientific communities dynamics. Our mission was to reproduce an analysis method based on co-references already implemented in python in a web tool allowing to visually explore the networks produced. A very high time constraint guides us to chose to develop this project under the form of a colocalized intensive workshop with the client. By navigating between ideas and constraints with the help of an agile method, we succeeded in producing simple yet efficient scientometric tool complying the scientific needs in a very short time.

A Data valorization project

  • Digital humanities
  • Visual analysis
  • Conception
  • JavaScript
  • React
  • TypeScript
  • Sigma.js
  • Graphology
  • Frist step: import CSV files data-set.
  • After parsing and indexation: filters settings.
  • Finally, the co-reference network with metadata nodes visualized.
  • Frist step: import CSV files data-set.

    Cross
  • After parsing and indexation: filters settings.

    Cross
  • Finally, the co-reference network with metadata nodes visualized.

    Cross

Toward Graph Commons v2

We join Graph Commons team to help out the development of the new version of their platform for mapping, analyzing, and sharing data-networks.. We work on refactoring the current client side code into React and update the Ruby-Neo4j stack. Our final goal is to add new features using the Graphology libraries allong with the integration of a new design currently under development.

A Consulting and support project

  • JavaScript
  • Graphology
  • React
  • Neo4j
  • Ruby
  • Data visualisation
  • DevOps
  • Architecture

You want to know more about what we do?

Learn more about our services