AGICOA is a non-profit organization that monitors the use of audiovisual works on cable or any other similar distribution method, collects royalties from operators, and distributes them to the rights owners.
Optimization of copyright distribution
AGICOA manages the collection and distribution of rights for over 1.6 million works by analyzing approximately 2.8 million broadcasts each year across 283 channels covering 48 countries. They have developed an in-house solution based on a set of rules, allowing an auto-identification rate of about 50% (associations not requiring manual verification) and processing about 400 broadcasts per minute. Artificial intelligence is used to match each broadcast to the correct work in their catalog, thereby improving the accuracy and efficiency of the process.
A relevant and targeted data collection
To ensure optimal matching, we gathered and structured all the key information from their catalog: titles, cast, production year, type of work, etc. This work was carried out in close collaboration with AGICOA's business experts to ensure that each selected data point met the actual needs of the process. Success Stories
Experience
The experience of the business experts was essential in identifying the relevant data and integrating specific rules refined over the years. At the same time, our artificial intelligence specialists analyzed different models and algorithms to optimize the accuracy and efficiency of the matching process.
Ambition
The goal of the project was to improve auto-identification by 30% in the matching process, reaching a rate of at least 65% and reducing processing time by 100%, achieving 800 broadcasts per minute. The new tool must also be scalable and capable of continuous learning, incorporating new identifications validated by business experts each year. Success Stories
Challenge
Given the heterogeneous nature of the broadcast sources, some suffer from encoding issues and varying quality, making identification more complex. Furthermore, the model needed to be capable of handling multiple languages (French, English, Finnish, Spanish, etc.) while accounting for cultural and linguistic specificities. Finally, the identification of series presented an additional challenge due to their growing volume and international franchises.
Solution
Our AI-based matching tool combines advanced NLP techniques to generate relevant suggestions and uses Machine Learning algorithms to identify the best match. This approach significantly improves the reliability and efficiency of the process, with the final result achieving an 80% auto-identification rate and 1,900 broadcasts per minute.
Our collaboration with Synchrotech on an artificial intelligence project allowed us to explore the IA potential and to benefit from their leading expertise in this field. Initially cautious, we were convinced by the outcome in accuracy, efficiency, reliability, and speed. Now being deployed, this project represents a significant step forward for our operations.
Annika Andrivet
Head of IT Department, AGICOA
Contact Us
Contact Us
By email: info@synchrotech.ch
By phone: +41 27 322 69 54
It's as simple as making a phone call 😊
Contact our CEO Gilles Fiorio directly
Come see us and have a coffee
SYNCHROTECH SA
Rue du Châble-Bet 41
CH-1920 Martigny
+41 27 322 69 54
SYNCHROTECH SA
Avenue de la Praille 31
CH-1227 Carouge, Genève
+41 22 552 20 28