How this Chinese AI champion began by helping the police solve crimes

This article is co-written by Li Tao.

When the Chinese government named a new batch of national champions in AI in August, one name was relatively unknown to most of the population.

MiningLamp, a Beijing based big data company, joined homegrown tech giants Huawei Technologies, JD.com, and Xiaomi Corp. as companies tasked with spearheading innovation efforts in the field.

Photo credit: Pixabay

Although not as well known as US equivalent Palantir Technologies, which reportedly contributed to America’s success in hunting down Osama bin Laden, MiningLamp’s data mining software is used to spot crime patterns, track drug dealers, and prevent human trafficking.

“Cases are being resolved on our platforms every day” in more than 60 cities and regions in China, said founder and CEO Wu Minghui. “We can run fast analysis on potential drug dealers or major suspects, improving the overall case-solving efficiency several hundred times.”

MiningLamp’s software enables users to search huge volumes of heterogeneous data – information with a great variety of types and formats – and process that into actionable knowledge and insight using a combination of proprietary and commercially available data management tools.

For example, police in far-flung cities may use different suspect descriptions and methods of recording evidence for theft cases. Using data mining, connections can quickly be found among the disparate data instead of having to manually cross-check dozens of case files.

Dressed in a white business shirt and sporting a fresh crew cut, Wu recounted to the Post how the decision to go into public security five years ago came from a childhood “pain point.”

While growing up in the seaside city of Yantai in China’s northeast Shandong province, Wu’s father worked as a policeman, dealing with grassroots level community problems such as noise complaints, local disturbances, and even quarrels among families.

“Growing up, I hardly saw my dad,” said Wu, in a quiet moment of contemplation. “Police are under a lot of pressure in China where the population per officer is about three times the world average.”

When pitching the software to early clients, Wu would even point out its advantages in helping small-town police officers spend more time with their children.

In training machines to think like humans, Wu’s engineers rely on the knowledge graph, a concept used by Google to enhance its search engine results with information gathered from a variety of sources.

“The human learning process also requires us to weave a network or database of knowledge and build connections through logical reasoning and associative thinking. Such an ability is often a determinant of one’s intelligence level,” he said.