Research background and purpose: Participatory budgeting (PB) plays an important role in a city live. It strengthens democracy, improves transparency of public expenditures, helps allocate resources more efficiently, and increases civic responsibility. In order to assess how such a mechanism contributes to public involvement and which initiatives are prioritized by the community engaged in the participatory budgeting process, a pilot study was undertaken for a selected Polish voivodeship city - Kielce. Attention was paid to the category of winning tasks, considering the classification by project type and its narrative. Design/methodology/approach: The exploratory analysis was carried out on real unit data. Data on projects submitted by residents were collected from websites using web scraping tools. The article relies on the case-study method as well as on the methods of quantitative and qualitative analysis. The narrative data were explored with the assistance of machine learning technology – Natural Language Processing (NLP) and the results were illustrated in the form of a Word Cloud. Findings: The conclusions of the picture of participatory budgeting in the chosen city are decidedly positive. It is a tool that promotes democratization of decision-making processes and involvement of citizens in public life. It is significant that the local community opts for soft projects, particularly green ones. Value added and limitations: This study investigates the changes in citizen engagement in PB using a time- and project-type-oriented quantitative approach. Furthermore, a machine learning-based qualitative approach in terms of NLP was employed to capture the major PB issues and topics, which is rather absent in this research field. As a pilot study, this research is limited to a single city and begins to explore text processing algorithms in such analysis. The promising results suggest potential for research development in the future.