Improving Public Services Accessibility Through Natural Language Processing: Challenges, Opportunities and Obstacles

The adoption of AI in the public sector processes and operations is already showing significant benefits, improving for example their efficiency and effectiveness in delivering services. In this direction, the EU-funded easyRights project explores the application of NLU techniques to improve service accessibility and in particular to extract from administrative documents an effective and step-wise description of the user experience. The project is devoted to easing the understanding of service procedures and improving the experience for the service users. The easyRights project especially aims at easing the access to public services to immigrants so targeting a special users’ category whose social fragility is further challenged by bureaucratic complexity. The present paper delineates the work done by applying NLU techniques to service-related administrative documents in four European cities. The first part describes the NLU system intended to play the role of a pathway generator and further outlines the basic architecture of the pathway composed of four key descriptors for each step, namely the “what”, “when”, “where”, and “how”. In the second part, the article discusses the initial outputs of the experiments related to a total of eight services developed in four pilot projects. Evidence of the key obstacles encountered is therefore discussed. Finally, the paper critically reflects on the general value of the achievements and findings and opens up some crucial questions related to the relevant lessons learnt from the side of the public authority and their preparedness to fully exploit the potential benefits from the adoption of AI solutions.