Article in the Same Section:	Hologram Stickers

The concept of artificial intelligence can now be applied to combat and prevent crimes, a research has found.

Professor Lam Kai Pui of the Department of Systems Engineering and Engineering Management of The Chinese University of Hong Kong, has developed the Artificial Intelligence Crime Analysis and Management System, supported by The CUHK donated funding of $150,000 in 1996 and 1997.

The research was conducted by The CUHK and the Hong Kong Police Force. The system was at first being tested in the Tuen Mun New Town police division and is being implemented in 52 police divisions.

The whole system consists of three parts. They are the Knowledge Database System, the Facial Identification Tool and the Map-based System.

The Knowledge Database System is supported with artificial intelligence.

It contains data about criminal investigations of experienced police officers and their analysis on the pattern, nature and method of different crimes.

“Take a car theft as an example. The officer just needs to insert data about the venue, time, technique, and devices used.

“The system would provide detailed analysis about the criminal, the pattern and suspects based on present case parameters and information in the database,” said Prof. Lam,

Besides, it would find the correlation with different crimes in terms of the criminals and methods used, thus fostering the investigation procedures. Though the Knowledge Database System is very useful, it will only be implemented in the Tuen Mun police division for the time being due to technical problems.

“Information stored in the system needs frequent updates to provide accurate analysis for investigations. This is the major technical difficulty for the system,” said Prof. Lam.

The pattern-analysis is another difficulty, as it is difficult to fix a pattern for different crimes which may be affected by many geographical and psychological factors.

It is also hard for experienced officers to describe their experiences clearly during data collection.

Hence this part of the system is still under research for further improvements.

The Facial Identification Tool consists of a facial feature database producing an accurate, infinite numbers of possible suspects’ images.

The image-producing procedure takes about 16 minutes and the system can produce images with 80 percent accuracy.

Police officers can produce a rough image according to the description of the witness or victim.

Having seen the rough image, the witness or victim can then fine-tune the characteristics of the suspect through the memory-recalling process.

The system then match the image with the suspects' photos stored in the database.

Hence, victims or witnesses do not need to view a huge amount of photos stored in the database. Hence, victims or witnesses do not need to view a huge amount of photos stored in the database.

Images produced can be stored for later reference.

If the patrolling officer is equipped with a notebook computer, the facial recognition process can be done immediately after the crime.

The Map-based System is comprised of a digitised map with names and locations of buildings, public infrastructures and different shops.

There would be information about the nature, statistics and sites of the crimes in different areas.

The system is mainly used in the allocation of manpower in patrolling.

Senior officers can allocate a suitable number of patrolling officers according to the number of crimes shown on the digitised map in different districts.

The Facial Identification Tool and the Map-based System are now implemented in 52 police divisions in Hong Kong.

An additional grant from the Research Grants Council was appropriated to Prof. Lam for further improvements on the system.

The system would potentially be implemented in other judiciary institutions like the Independent Commission against Corruption, the Customs and Excise Department and mainland police units.

The system increases the efficiencies in crime prevention and investigation