Large-scale Alarm Police Log Semantic-spatio-temporal Mode Interactive Exploration

The alarm log contains both time,space,and case description information,and is a typical unstructured spatio-temporal data.Compared with spatiotemporal social media,there is less connection between the data items of the alarm log,and no complicated network relationship is provided to provide valuable clues for mining data patterns.Therefore,its analysis relies more on the mining and semantics of rich semantics.Exploration of time and space patterns.The interactive exploration of large-scale unstructured spatio-temporal data is a challenge,and the existing research focuses on structured spatio-temporal data.In response to this problem,this paper proposes a visual analysis framework to support the interactive exploration of spatio-temporal patterns of large-scale unstructured alarm logs.First of all,this paper proposes a method based on topic model integration to extract topics from heterogeneous texts to avoid the problem that traditional topic models are difficult to determine reasonable parameters.Secondly,the framework contains a data cube whose three dimensions are time,space and theme.It is used to pre-store semantic indicators in different space-time coordinates,so that it can quickly respond to users' requests for different semantic-spatio-temporal content queries.Thirdly,based on this data volume,this paper designs and implements a visual interactive system to support historical data aggregation analysis and semantic spatio-temporal pattern prediction.Fourth,based on this data volume,this paper designs and implements a visual interactive system to support historical data aggregation analysis and semantic spatiotemporal pattern prediction.We use the real alarm log of a certain city in China to verify this aspect.The discovered patterns and various performance indicators verify the effectiveness of the method.