Successful cases
5WAI, an AI system to evaluate the journalistic quality of the news stories:
Goienako albisteen kazetaritza-kalitatea neurtzeko adimen artifizialeko sistema bat garatu da. Sistemaren helburua da albisteetan kazetaritzaren oinarrizko 5W galderak —zer, nor, noiz, non eta zergatik— automatikoki identifikatzea, argitaratu aurretik edukia osatua den egiaztatzeko eta artxibo-funtzioetarako informazio egituratua sortzeko.
5WAI, an AI system to evaluate the journalistic quality of the news stories:
Overview
We have developed an artificial intelligence system to evaluate the journalistic quality of the news stories of the Basque-language communications group Goiena. The aim of the system is to automatically identify the 5 Ws —what, who, when, where and why— in the news stories, to verify that the information is complete before publication and to generate structured information for archival functions.
Challenge
We identified the need to automate the verification of the 5 Ws, as doing this manually entails a significant amount of work. On a technical level, the biggest challenge was that the answers to these questions are not always explicit, nor appear at a single point in the text. The information is often dispersed or expressed in an abstract way, thus requiring a great capacity for understanding the language.
Collaboration
It was developed in collaboration with the communications group Goiena.
Result
We created a technology service capable of extracting and analysing the 5 Ws, by adapting and evaluating LLMs that would be able to identify the 5 Ws. Furthermore, we developed an API and a demo interface, allowing users to send a news story and receive information about the 5 Ws. Goiena can therefore use this automated tool to improve the quality of its news stories and enrich the searches in its archive.
