Successful cases
System for automatically tagging Naiz.eus content:
We have integrated an automatic semantic tagging system for news and written content into the Naiz.eus digital communication environment. This system analyses the content of the news and automatically proposes thematic descriptors, keywords and relevant entities, to facilitate content organisation, search, relevance and personalisation.
System for automatically tagging Naiz.eus content:
Overview
We have integrated an automatic semantic tagging system for news and written content into the Naiz.eus digital communication environment. This system analyses the content of the news and automatically proposes thematic descriptors, keywords and relevant entities, to facilitate content organisation, search, relevance and personalisation.
Challenge
The main challenge was to automate the semantic annotation work performed manually by the journalists, while maintaining adequate quality and reducing, insofar as possible, the need for manual review. For this, it was necessary to define Naiz’ category system, prepare a corpus of annotated news articles, adapt the neural models to the domain and build an agile service capable of integrating the system into the daily CMS workflow.
Collaboration
It was developed in collaboration with NAIZ.
Result
We developed an integrated system for the automatic tagging of Naiz.eus content. For each news article the system proposes the main themes, keywords and relevant entities to facilitate content organisation, search, personalisation and analysis. For this, linguistic models and autoencoder-based LLMs to extract thematic descriptors, specific descriptors and relevant entities. The solution was integrated into the CMS workflow through API, enabling the semantic tagging to be carried out more quickly, systematically and scalably.
