About this Research Topic
Due to the nature of randomness, predicting the non-recurrent emergences of most spatiotemporal events are prohibitively difficult. Estimating the future impact of spatiotemporal events, on the other hand, addresses a heuristic problem with the chain of causality, and brings invaluable research opportunities in understanding the cause, formulating resilience, and planning the exploitation of the events. Although many believe that the occurrences of the social events are neutral, their impacts on human society, on the other hand, can be either favorable or deleterious. For example, the temporal duration and spatial congestion caused by traffic incidents consequence enormous socio-economical losses every day. While the emergences of the underrated disruptive technologies bring in millions of profits for venture capitalists. Therefore, to foresee and quantify the level of damage the deleterious events will charge and the amount of profit the favorable events will earn has great research values and should hit the spot of everyone’s interest.
The most recent decades have witnessed the prominent performance of machine learning models in a wide variety of tasks such as feature learning, classification, and pattern recognition. With the abundance of data generated by the ubiquitously deployed sensors (i.e., traffic speed sensors, climate sensor networks, social networks, political polls, etc.), we believe that there will be tremendous research opportunities for machine learning models in estimating the future impacts of social events. Differing from anomaly detection in the pattern mining field, social events mining and impact estimation address the problems of learning the representations of the events from multiple heterogeneous data sources (i.e., social media, urban sensors, news articles, and research publications) and forecasting the future socio-economical influence from multiple aspects. In the scope of estimating the social impacts of spatiotemporal events with machine learning models, we announce four major target directions of this Research Topic: 1) learning the representations of events, 2) quantifying the impacts of events, 3) modeling the patterns of events, and 4) forecasting the trends of events.
Submissions may include but are not limited to:
- Learning the representation of events: models that can extract representative features from fusions of heterogeneous sensor datasets and formulate the definition of social events
- Quantifying the impacts of events: models that can quantify the social impacts in specific target domains. Such impacts include temporal duration, spatial congestion, and topical influences
- Modeling the patterns of events: models that can recognize the learned patterns from heterogeneous data sources and generalize the patterns in other domains
- Forecasting trends of the events: models that predict the future trends of the social events given a combination of heterogeneous sensor data, and inspire further studies
The most recent decades have witnessed the prominent performance of machine learning models in a wide variety of tasks such as feature learning, classification, and pattern recognition. With the abundance of data generated by the ubiquitously deployed sensors (i.e., traffic speed sensors, climate sensor networks, social networks, political polls, etc.), we believe that there will be tremendous research opportunities for machine learning models in estimating the future impacts of social events. Differing from anomaly detection in the pattern mining field, social events mining and impact estimation address the problems of learning the representations of the events from multiple heterogeneous data sources (i.e., social media, urban sensors, news articles, and research publications) and forecasting the future socio-economical influence from multiple aspects. In the scope of estimating the social impacts of spatiotemporal events with machine learning models, we announce four major target directions of this Research Topic: 1) learning the representations of events, 2) quantifying the impacts of events, 3) modeling the patterns of events, and 4) forecasting the trends of events.
Submissions may include but are not limited to:
- Learning the representation of events: models that can extract representative features from fusions of heterogeneous sensor datasets and formulate the definition of social events
- Quantifying the impacts of events: models that can quantify the social impacts in specific target domains. Such impacts include temporal duration, spatial congestion, and topical influences
- Modeling the patterns of events: models that can recognize the learned patterns from heterogeneous data sources and generalize the patterns in other domains
- Forecasting trends of the events: models that predict the future trends of the social events given a combination of heterogeneous sensor data, and inspire further studies
Keywords: Artificial intelligence, machine learning, forecasting, spatiotemporal
Important Note: All contributions to this Research Topic must be within the scope of the section and journal to which they are submitted, as defined in their mission statements. Frontiers reserves the right to guide an out-of-scope manuscript to a more suitable section or journal at any stage of peer review.
