Smart policing involves public security departments putting people first. This means comprehensively controlling surrounding elements such as houses, vehicles, roads, networks, sites, and organizations, in a concerted attempt to establish a dynamic trajectory management and control mechanism (Hardy III et al., 2024). These elements include events, locations, objects, organizations, virtual identities, attribution objection, spatial-temporal relations, semantic relations, and feature relations—all of which constitute knowledge inherent in the existing intelligence analysis system (Keppens et al., 2006). Additionally, customs anti-smuggling departments need business information such as capital flow, goods flow, and document flow to better perform information management.
Regarding information element extraction, public security information mainly consists of semi-unstructured case information and unstructured document information entered into a database. Early information extraction methods were mainly based on the recognition of rules, which achieved recognition and extraction of information elements by establishing the rule module of information elements. Extraction methods based on Chinese word segmentation have been increasingly utilized, using statistics-based and dictionary-based word segmentation methods. In recent years, extraction methods based on machine-learning have been applied, for example, conditional random fields (CRF), support vector machine, hidden Markov models, and conventional neural networks (Raffaele et al., 2021).
The model combining convolutional neural network and conditional random fields can be applied to entity recognition and information extraction; the use of convolutional neural networks, combined with conditional random fields, can effectively complete the extraction of text information. Fusing this with structured data information related to anti-smuggling case events, such as customs system and public security systems, can encourage multi-level and multi-granularity semantic integration of anti-smuggling case event information. Table 1 shows the related event text information.
Table 1. Sample extraction of textual information on smuggling cases
| Cases | Time Information | Spatial information | People | Goods |
|---|
Shenzhen customs seizes 863 “live animals”
| December 10, XX and from mid-November to December 4, 2019
| Vietnam, Laos, Philippines
| Shenzhen passenger
| Live corals, giant clams, live turtles, meerkats, monkeys, crocodiles, otters
|
Macau customs seizes millions of dried Vietnamese mushrooms
| December 16, XX, October-November 2019, during the African Swine Fever program
| Vietnam
| Macau passenger
| Dried shiitake
|
| China and Japan join forces to crack 500 kilograms of methamphetamine smuggling case | December 12, XX, December 11, XX | Off Fukuoka Prefecture, Kyushu City, Vietnam | Active offenders, accomplices, Chinese, Japanese, Vietnamese | Methamphetamine |