A multilingual, machine learning platform for aspect-based sentiment and emotion analysis

01 January 2021 → 31 December 2022
Regional and community funding: IWT/VLAIO
Research disciplines
  • Humanities
    • Computational linguistics
aspect-based sentiment analysis emotion detection natural language processing machine learning
Project description

In this project, we aim to develop a fine-grained aspect-based sentiment and emotion analysis architecture, which is multilingual (English, Dutch, French and German) and which is completely data-driven. The system should also be adaptable: users should be able to correct the output of the system on their data as the basis for retraining the system on the corrected data, leading to a more qualitative output which becomes increasingly tailored to company-specific data. The prototype will also be accompanied by two dashboards: a dashboard which provides a comprehensive visulatisation of the data and an annotation dashboard which allows to correct data and retrain the machine learning architecture.