Business Case

Radio France MetaData

Reconcile heterogeneous data to serve the business.

MediaData is the documentary service’s project. It onboards the Perfect Memory Platform to reconcile and enrich contents located in several specifics data sources.
This project built a solid base for the information reconciliation and propagation. Perfect Memory contributed to the specification based on Radio France use cases.
The fruit of this work is available via a PoC platform with thousands of contents of France Inter, France Info and France Bleu. It demonstrates how the Perfect Memory platform can solve use cases to the business teams thanks to the reciliation of heterogeneous data coming from the following systems:
  • EDITO (channel director)
  • ITEMA (On Air director)
  • EDISON (thesauri)
  • Digital management Web database for the indexation
  • ORPHEA, (picture integration) 
  • NETIA logs for the synchronisation with the on air episode

A few numbers…

TV channels
Radio stations
0 m
listening on demand

Key goals


Simplify content reutilization


Provide a rich and business oriented portal for search, consultation and exploitation


Improve the user experience


Make exchange more fluid


Maximise content exposition


Centralized and immediate access to content

Bring data back together

The MAM-as-a-Brain is fed by any data available from the field.

Various types of data coming from various sources are reconciled and transformed into  meaningful information, understandable and usable by the business teams. 

Metadata for On Air management, web publication information, diffusion logs…. All this  data variety be centered on content through the reconciliation base defined by the  business team language. 

Content are automatically  indexed and are available for production and  archivists with reconciled information, increasing the access point to get those information. 

As an example, the topics detected within the on Air director titles  enrich the programme’s with the related information.

Centralized and immediate access to content

All enriched and collected content are immediately available for business teams in our  graphical user interface Trouvé™, ergonomics designed for operational efficiency ... and  user enjoyment 

The search engine provides dynamic facets and filters which are fed according to the data’s  diversity reconciled from the various production, distribution and publication systems, plus  the automatic extraction tools. Thus, the search is always providing more and more  available information to give the business teams more opportunities to find content.

Highly indexation of the audio recording

The collaboration of extraction services helps reveal relevant information within the content.  

Persons, locations, organisations… The automatic speech transcription coupled with the named entities recognition clarifies the information allowing navigation and a rich and contextualized intra program search.

Navigate between occurrences of the same entity, provide a report of the  most cited entities, filter and cross criteria …
The exploitation of a specific  portion of audio content is immediate. 

The speakers relevant example

MediaData project demonstrates the capacity to make content searchable and accessible.  

In particular, each sentence extracted from the content becomes a searchable  element according to the expected business criteria: gender and identity of the speaker,  program… 

Also, speaker identification is the perfect example of the benefits of reconciliation of information. Indeed, this is not obtained by the mobilization of a trained extraction tool, but by the deduction of cross-information according to identified business rules

Therefore, the combination of information regarding an interviewee (coming from the web) and information about the most frequent speaker within a content (coming from the transcription) makes possible the identification of speakers for every sentence.  

This is how the access to a specific sentence becomes immediate.

Interconnect audio, video and image content

The broadcasted audio programs are automatically connected to related  multimedia content, through their common information. 

Therefore, the photos of the guests stored in the Orphéa system are  accessible directly from the navigation of the information relating to a  program. 

The discovery of connected elements improves the  user navigation, enhances all the content produced  by the house of Radio France, in a shared and harmonized manner.

A rich indexation environment

The user interface is also designed for content indexation, the integration of  Radio France dictionaries and thesauri (genres, broadcasts, etc.) and shared  standards (IPTC, IAB categories, etc.). 

The granularity of the documentation is multiple and is consistent with the  business needs of annotations on the whole content or on specific  portions of it.

Improve exchange with production teams

Users have personal folders allowing them to save extracts to share.  MAM-as-a-Brain allows you to export any extract to connected production tools. 

In all circumstances, allowing the use of any content from any period shifts the logic of archiving to asset management.

Business KPIs

Viewer@3xCreated with Sketch.

Consistency and centralization of 6 non communicating data sources

3 business-oriented automatic extraction tools collaborates 

6 months of Inter, Info and Culture content, 2,000 episodes, 150,000 indexed  sentences,  160,000 referenced entities 

Viewer@3xCreated with Sketch.

Validation of the agility of the solution through the iterative process of taking into account business needs over 4 months of discussions


Data is interpreted to be delivered as exploitable assets and meet the needs of your operational context.


Increase of users’ independence and reactivity. Users can access any rich, complex and meaningful content on their own.


Unleash data and content governance (legal, business, operational). 100% of the data and content are indexed and monitored.

Want to know more?
Meet one of our experts!

PERFECT MEMORY is a Software Publisher. We crafted a new generation of Digital Asset Management Systems, the DAM-as-a-Brain, which can refine and execute cognitive processing on content. 

Our DAM-as-a-Brain allows to collect, interpret, and transform any data and content into natively exploitable assets for operational teams and their ecosystems. 

We guarantee to our users a quick, intuitive, and exhaustive access to data and content, regardless of their nature and origin.