Semantic Understanding Of Meteorological Disaster Events Based On Formal Concept Analysis

In the context of the global warming climate,meteorological disasters occur frequently in local areas,which attracts more and more attention.Accurate disaster information is critical to disaster response.However,observation data such as traditional meteorological disasters have data sparsity?hysteresis?and high cost characteristics.In the social perception,each resident individual plays the role of sensor,Using the characteristics of high-resolution and ubiquitousness of social perception data such as Weibo,it provides a new perspective for exploring the theme of event and the temporal and spatial behavior patterns of meteorological disasters.This paper takes "Meteorological disaster knowledge mapMeteorological disaster ontology library construction-Disaster event topic detection" as the main line of research,and then carries out the hotspot and trend analysis of meteorological disaster research based on knowledge map method;The meteorological disaster ontology is constructed using the current national standards for meteorological disasters and meteorological industry standards,Finally,the formal concept analysis(FCA)method is used to extract the social media data collected during a rainstorm weather process,which provide public support for effectively understanding the disaster relief information and public attention hotspots.The main research contents are as follows:(1)The paper explores the frontiers of meteorological disasters through knowledge maps,and finds that the perceptual extraction of residents under sudden meteorological disasters is one of the hot research directions.Through the formal description of meteorological disaster standards,the ontology of meteorological disasters is constructed,which helps computer to classify and understand meteorological disasters.Subjective detection of social media texts can monitor residents' concerns about hotspots and emotions in disaster weather conditions in real time,and realize the perception and feedback of residents under meteorological disasters.(2)Firstly,based on the keyword clustering and evolution analysis of knowledge maps,this paper finds that precipitation,flood and drought are some of the most influential meteorological disasters.In recent years,emerging observation techniques and data sources,such as remote sensing and social big data,have also provided new ideas for meteorological disaster research.Meteorological disaster early warning and meteorological disaster emergency management are also new research hotspots.(3)Secondly,this study combines the current national meteorological disaster standard GB/T 28921-2012 analysis,uses protege to sort out meteorological disasters and classification,data attributes,disaster attributes,uses OWL language for ontology description,and uses Hermit inference machine to complete ontology correlation verification,completed the construction of meteorological disaster ontology.(4)Finally,using the Weibo big data during a heavy rain in Hefei as the data source,the formal concept analysis method is used to carry out the topic extraction of the Weibo text in the rainstorm process,and the final form concept result is reduced by threshold calculation and stability selection.The results show that FCA is an effective topic extraction method,which can be used to detect event topics under different threshold granularities according to different system tasks.