You are reading it here first: In a move that’s supposed to boost predictive policing but can lead to profiling of historically marginalised and vulnerable groups, the Tamil Nadu government has decided to conduct geographic information system (GIS) mapping of crime zones under the Chennai Safe City Project.
The Greater Chennai Police, which has floated a tender for selecting a system integrator for constructing a GIS Centre and conducting GIS mapping, has said that the selected entity would —
- Conduct historic crime mapping for 5 years, from 2016 to 2020, on cases related to crime against women
- This data will help in analysing “hotspots/emerging hotspots’ and spatially identifying different kind of crime trends based on various parameters
- The GIS platform will also keep an eye on social media for getting a “macro-level analysis of given location” based on keywords.
- Locations of CCTVs in Chennai will also have to be mapped into the system. From there, a police official can be able to view the footage of a CCTV by linking the system to the video management system (VMS).
Agencies worldwide are using geospatial technologies for mapping crime, identifying crime hot spots, assigning officers, and profiling offenders. Spatial analysis is giving geographical context to real world incidents and helping police officials to create geographical profiling of offenders — Greater Chennai Police in the tender
Studies by researchers have shed light on the inherent biases in GIS mapping of crime zones and how they can end up targeting underprivileged sections of society. It is important to point out that that this flawed technology is being introduced under the Safe City Project. Earlier, MediaNama had exclusively reported on how the Uttar Pradesh government was introducing drones, facial recognition, and video analytics in the guise of keeping women safe.
Read MediaNama’s in-depth coverage of the Lucknow Safe City Project here.
A look at what Chennai Police wants from the GIS mapping system
“The envisaged Women and Child safety-centric GIS solution will include setup of a GIS center, with requisite hardware and software, application & licenses of the GIS/BI/Analytics tools, training of staff, and operation and maintenance for three years,” the tender reviewed by MediaNama read. Apart from that, the Greater Chennai Police said that the system integrator should —
- Deploy a solution that institutionalises data collection and integration, maps all infrastructure and assets related to women’s safety
- Build and deploy predictive analytics and BI/MIS tools to visualise the data
- Integrate the system with base maps that are accurate up to 0.5 meters
The mapping system will draw its data from —
- Crime and Criminal Tracking Network & Systems (CCTNS) data such as latitude/longitude, type of crime, perpetrator details and relevant data fields.
- Community Service Register
- Obtain complaint information and location details from Dial 100/112
- Integration with Tamil Nadu’s Kaavalan App to crowdsource information
- IoT devices for obtaining information with regards to camera failures, alerts via surveillance project, Smart City sensors for disaster mitigation, etc
- 181 Call Centre data integration
What about its security?
The System Integrator must adopt an end-to-end security model that protects data and the infrastructure from malicious attacks, theft, natural disasters etc. Successful Bidder must make provisions for the protection of the software system from hackers and other threats. Using Firewalls, DDoS and Intrusion Prevention Systems such attacks and theft should be controlled and well supported (and implemented) with the security policy as per MoHUA guidelines — the tender read
The Greater Chennai Police has laid down guidelines for the system integrator to follow in terms of security practices. They are —
- The selected SI will have to build a trail of all activities and operations using log reports so that errors in the system – intentional or otherwise – can be traced and corrected.
- Logs should be archived for future analysis and forensics
- Authorities would have to conduct a security audit within 3 months of operationalisation of the system
- There must be access controls to ensure that the system is not tampered with or modified by the system operators or unauthorised persons
Interestingly, the Greater Chennai Police has proposed a penalty for any security breach for the selected vendor. “For every solution Security-breach/Vulnerability attack with severity level of 1, 2, 3 & 4 (as defined in section 22.214.171.124) then a penalty of 2%, 1.5%, 1%, and 0.5% of OPEX respectively and in case of similar attacks of more than three attacks may lead to termination of contract,” the tender read.
Why does Chennai Police need a GIS crime mapping system?
A series of use cases was provided by the Greater Chennai Police as a reason for including hospitals and old age homes as different ‘layers’ for the proposed crime mapping system.
Besides what's included in the table, the Chennai Police reasoned that the crime mapping system will —
- Identify suspicious movements/events/incidents and aiding in investigation
- Support pattern and trend analysis in different police zones
- Forecast crime incidents
- Identify patterns
- Monitor sex offenders who have been released
- Monitor offenders' travel pattern
- Identify linkages among criminals
A look at Delhi's crime mapping system and its inherent biases
Similar to the crime mapping system proposed by the Greater Chennai Police, Delhi Police's Crime Mapping, Analytics and Predictive System (CMAPS) was developed by the Indian Space Research Organisation (ISRO) to help identify and map crime patterns. Electronic data, such as addresses of victims, that Delhi Police receive through its single emergency helpline number 112 is retained forever, SK Singh, the DCP (Operations and Communications) of Delhi Police had told MediaNama in 2020.
The system also used big data analytics. For instance, crime data around Dussehra helps to decide where to put pickets. “The crime pattern remains the same. If Ram Lila is happening at Ram Lila ground, so it will happen this year also and the crime had happened last year, the criminals will also target that area,” Singh had said.
Researchers Vidushi Marda and Shivangi Narayan studied Delhi's CMAPS for over 2 years (from February 2017 - March 2019). In their report 'Data in New Delhi's Predictive Policing System', the researchers found a three-fold bias in the system.
- Under-representation of individuals from upscale areas: The study found that input data from individuals from upscale areas of Delhi did not feature as much when compared to data of those from, for instance, slums.
The DMD receives around 20,000 calls a day, and in the course of our research some employees said that people from posh areas “hardly called”, and that an overwhelming majority of these calls were from slums. This means that the probability of crime will be marked higher in hotspot areas where quantity of engagement is higher, leading to a vicious circle of heightened scrutiny for the most marginalised, eventually leading to more arrests and reports coming out of these areas — excerpt from Data in New Delhi's Predictive Policing System
- Historical and human bias: The study argued that the act of gathering information has always been a selective one; "with greater surveillance often befalling axes of disadvantage, i.e. caste, gender, class and religious minority". Not just that, the researchers found that another complication in the form of a human bias is introduced when someone is tasked with choosing which area or under which crime a certain call should be filed.
- Measurement bias: The study found that there was lesser scrutiny in privileged neighbourhoods when compared to temporary settlements due to the difference in the spatial accuracy between the sets of places.
Given that the spatial distribution of Delhi is less accurate among temporary settlements, and there is greater nuance in data arising from privileged neighbourhoods in Delhi, the clusters of information tend to be less quantitatively overwhelming, thus attracting less future scrutiny. This bias arises not just because of systemic blind spots, but also because of vulnerable individual’s inability to engage with the system as well as others — excerpt from Data in New Delhi's Predictive Policing System
Chennai system will subject particular areas to more surveillance
Shweta Mohandas, policy officer at the Centre for Internet and Society (CIS), said that the Chennai mapping system will increase surveillance in areas where the police perceive crime to be high. "This would again increase the crime data, thereby subjecting the area and its people to more and more surveillance," Mohandas said.
There needs to be clarity on what are the crimes against women that will be mapped, since there are a number of crimes that are not specific to women alone and also that crimes against women are often committed inside closed doors — Shweta Mohandas
She also pointed out that the tender specifically mentions that this technology may be used on 'suspects' or persons of interest. "One of the concerns is that the (policy/tender) does not specify whether only the records of the convicted people will be used, it is worrying if the records of accused people are used as some might have been acquitted, but would still be in the database as persons of interest," Mohandas said.
- How Delhi Police uses data from emergency calls to track, map crime
- Geospatial guidelines will help Indian mapping companies compete globally, govt says
- Government restricts foreign mapping services to one metre accuracy
- Targeted procurement, outsourcing, AI/ML: How Survey of India is planning digital mapping of villages using drones
Have something to add? Post your comment and gift someone a MediaNama subscription.