Elhuyar – Real-time analysis of the digital reputation of companies and institutions, drawing on extensive media coverage
Elhuyar
Sector: Services
Business Case
The motivation behind systems for analysing the reputation of companies and institutions in print media lies in the need for organisations to manage and protect their public image, which is crucial for long-term success. These systems make it possible to continuously monitor and evaluate how companies are perceived by the media.
Objectives
Automate real-time analysis of the digital reputation of companies and institutions, drawing on extensive media coverage. The analysis to be automated includes the identification of the polarisation concealed behind such opinions or news, together with the automatic measurement of reputational attributes of different dimensions in order to compare them with those of the competition.
Use case
Development of a module for polarity classification and reputational attributes at the entity level in the context of a news story using both encoder and decoder type neural language models.
Infrastructure
Datasets privados de noticias que incluyen anotaciones de polaridad y atributos reputacionales a nivel de entidad.
Technology
Automatic or deep learning Text Mining Voice recognition
Data
Private news datasets that include polarity annotations and entity-level reputational attributes.
Resources
Researchers who specialise in NLP and especially in Information Extraction and LLMs. Server infrastructure to deploy the trained models.
Difficulties and learning
Difficulty of training models without establishing extremely clear criteria to define the different reputational attributes to be identified in the process, so the process of annotation and definition of criteria was reinforced, guaranteeing an inter-agreement threshold.
KPIs (business impact and metrics of the model)
Success rates above 85% in the analysis of polarity and classification of reputational attributes. Processing of up to 25,000 news items per month per user. Automation of the reputation management process with capabilities equivalent to manual execution.
Funding
Mixed financing of public grants and private investments.
Collaborators, Partners
GNOSS
