Feedback2Event: Public attention event extraction from spontaneous data for urban management
Александр Александрович Антонов, Georgii I. Kontsevik, Maksim Natykin, Sergey A. Mityagin · Procedia Computer Science · 2023
With the widespread use of social networks, citizens actively use this channel to discuss problems and events arising during life in the city. Thus, representatives of the authorities and other stakeholders can monitor the process of interaction between people and urban infrastructure and services in real time. However, monitoring is complicated by the noisy and unstructured nature of spontaneous data. Informed decision-making based on such data requires extracting the cause of the messages, formulated as an event, to infer them from the unknown and characterize them. The method proposed in the work allows to extract from the flow of spontaneous data events and their spatial and temporal characteristics, as well as to form connections between them. Results of the experiment shows that this approach can represent processes with semantic similarity in the context of urban spatial model with different levels of localization and variants of distribution. Such events can be used for the effective management of urban processes based on spontaneous data, which acquires for maximum permissible level of spatial detail.