Using NLP to classify terabytes of media files

E-learning media retrievable thanks to curriculum mapping

Our AI back-end services handle natural language, machine learning, and deep learning - in a GDPR-compliant manner.

confident girl at at notebook in a classroom

By integrating our module with the FWU – Das Medieninstitut der Länder platform and optimizing the user experience, media is automatically linked to curricula. Relevant metadata is extracted using natural language processing (NLP) and requires only final review and verification. The fact that relevant and appropriate media is significantly more discoverable meaningfully improves teaching and learning outcomes.

1,500

automatically captured curriculum chapters

93%

initial accuracy of topic classification

85%

initial accuracy of curricula affiliation classification

The challenge

E-learning content overflow overwhelms classification systems, hindering efficiency

During the pandemic, the amount of e-learning content available increased rapidly. Search engines and recommendation systems present at most schools or in existing e-learning systems were overwhelmed because they relied on the manual classification of added content. This is no longer feasible given the volumes. Currently, finding relevant content is very time-consuming. Because of this, publicly available content cannot be used efficiently.

Mundo-Screenshot

The solution

AI-driven platform enhances learning experience with enriched curricula and discoverability

We designed and developed a platform that presents school curricula in a structured way. Using NLP, each curriculum paragraph is enriched with topic- and skill-specific keyword metadata. One algorithm classifies the topics of generic input texts, and another matches this content to link it to the learning plans to make it discoverable. Our artificial intelligence (AI) back-end services use machine learning and deep learning to process natural language. Existing ontologies and taxonomies are efficiently incorporated. In short, this means that we throw in the content, enrich it with metadata, and the topic can be determined based on that. This makes learning fun.

Our partner

FWU

Mundo (http://mundo.schule) is a freely accessible media library. The SODIX platform aggregates educational media for learners, teachers, and educators. SODIX is operated by the FWU Institut für Film und Bild in Wissenschaft und Unterricht gemeinnützige GmbH - Das Medieninstitut der Länder (FWU). FWU has been producing audiovisual media as teaching and learning aids in education, upbringing, and science for more than 70 years.

Make the most of your data with AI.

Do you want to to classify terabytes of files and content?

Loading HubSpot form...

More of our achievements