Data mining and value-added analysis.
Law enforcement agencies, particularly in view of the current emphasis on terrorism, increasingly face the challenge of sorting through large amounts of information needed to help them make informed decisions and successfully fulfill their missions. At the same time, resources, particularly personnel, often dwindle. Described by one agency as the "volume challenge," (1) local, state, and federal agencies alike all struggle with an ever-increasing amount of information that far exceeds their ability to effectively analyze it in a timely fashion.
However, while these issues have surfaced, an extremely powerful tool has emerged from the business community. This tool, used by mortgage brokers to determine credit risk, local supermarkets to ascertain how to strategically stock their shelves, and Internet retailers to facilitate sales, also can benefit law enforcement personnel. Commonly known as data mining, this powerful tool can help investigators to effectively and efficiently perform such tasks as the analysis of crime and intelligence data. (2) Fortunately, because of recent developments in data mining, they do not have to possess technical proficiency to use this tool, only expertise in their respective subject matter.
WHAT IS DATA MINING?
Data mining serves as an automated tool that uses multiple advanced computational techniques, including artificial intelligence (the use of computers to perform logical functions), to fully explore and characterize large data sets involving one or more data sources, identifying significant, recognizable patterns, trends, and relationships not easily detected through traditional analytical techniques alone. (3) This information then may help with various purposes, such as the prediction of future events or behaviors.
Domain experts, or those with expertise in their respective fields, must determine if information obtained through data mining holds value. For example, a strong relationship between the time of day and a series of robberies would prove valuable to a law enforcement officer with expertise in the investigation pertaining to this information. On the other hand, if investigators, while reviewing historical homicide data, noticed that victims normally possessed lip balm, they would not, of course, associate lip balm or chapped lips with an increased risk for death.
WHY USE DATA MINING IN LAW ENFORCEMENT?
The staggering increase in the volume of information now flooding into the law enforcement community requires the use of more advanced analytical methods. Because data-mining software now proves userfriendly, personal-computer based, and, thus, affordable, law enforcement agencies at all levels can use it to help effectively handle this increased flow of data.
The law enforcement community can use data mining to effectively analyze information contained in many large data sets, even those involving written narratives (which represent a great deal of valuable law enforcement information). These may include calls for service data, crime or incident reports, witness statements, suspect interviews, tip information, telephone toll analysis, or Internet activity--almost any information that law enforcement professionals encounter in the course of their work. (4)
Not only can these data sets differ by type but they can originate from different sources, (5) potentially giving law enforcement agencies both a more complete informational base from which to draw conclusions and the ability to identify related information in separate databases or investigations. For example, this may prove valuable in the area of illegal narcotics enforcement. The law enforcement community frequently gathers information regarding markets, trends, and patterns, while medical and social services personnel store information concerning substance use and abuse on the individual level. In instances where appropriate, the opportunity to combine these data resources can give investigators a more complete picture and can help address various narcotics problems more rapidly, potentially saving both lives and resources.
Law enforcement agencies can consider exploring the use of data-mining applications to assist them in a variety of areas. Some examples include tactical crime analysis, deployment, risk assessment, behavioral analysis, DNA analysis, homeland security, and Internet/infrastructure protection.
Tactical Crime Analysis
Data mining offers law enforcement agencies potential benefits in the area of tactical crime analysis. For example, because agencies can use data mining for such purposes as to more quickly and effectively identify relationships and similarities between crimes and to forecast future events based on historical behavioral patterns, they can develop investigative leads and effective action plans more rapidly. (6) Major case investigations, which frequently present not only large volumes of information but also demands for rapid case resolution, serve as good examples of how law enforcement agencies can benefit from data mining in this regard.
Law enforcement agencies can use data-mining technology to help them deploy their resources, including personnel, more effectively and proactively. For instance, data mining can help them identify such key elements in a case or series of events as patterns of time and location--by forecasting future events based on this historical data, agencies potentially could anticipate strategic locations for deployment.
Data mining also allows agencies to consider multiple variables at one time and to add more weight to those considered most important to the decision at hand. For example, patrol officers, who generally respond to incidents with quick turnaround rates, may answer to numerous calls for service and effect many arrests in a relatively short amount of time. On the other hand, death investigations can require multiple officers' entire shifts just to maintain the crime scene perimeter; as a result, homicide investigators generally may handle considerably fewer incidents and arrests. To this end, by weighing heavily such factors as the type and duration of these incidents, law enforcement agencies can develop effective deployment strategies.
By using data mining, law enforcement personnel, for purposes of analysis, also can link incidents, crimes, or changes in crime trends to other types of events in making deployment decisions. For example, an agency historically may have noticed relationships between major weather events, such as snowstorms or hurricanes, and decreases in street crimes. Also, they may have seen how the arrests of key players in organized crime or drug distribution rings seem to result in increased violence as informants are sought and identified and as new leaders emerge during reorganization. As another example, they may associate increased apprehension rates and a strong economy with decreases in property crimes. (7) By using data mining to consider such relationships, law enforcement agencies then can deploy their personnel as they deem necessary.
Much like lenders and credit companies use data mining to great effect in assessing the financial gamble involved with lending money or extending credit to individuals or groups, law enforcement agencies can use it to characterize the risk involved in various incidents. For example, agency personnel can explore the use of data mining to identify common characteristics of armed robberies that ended in assaults; doing so then can help identify those that may escalate into assaults in the future. Similarly, in the past, certain types of property crimes have proven related to subsequent stranger rapes. (8) The ability to characterize property crimes as similar to those previously associated with subsequent sexual assaults can alert investigators to focus on certain cases and develop effective action plans, perhaps preventing many similar situations from occurring in the future.
The behavioral analysis of violent crime represents another area with significant potential for data mining. For instance, law enforcement agencies can use data mining to identify common behavioral characteristics in different cases. Even when not identifying a specific offender, investigators may find it possible to gain some insight into what type of offender may prove related to a particular incident. Research in this area, for example, has resulted in the use of data mining to efficiently link serious sexual assault cases based on similar offender behaviors. (9)
Law enforcement agencies also can benefit from the use of data mining when examining DNA evidence. For example, when DNA links a new suspect to an old case, investigators logically may wonder what other cases the suspect may be linked to. Given the amount of information involved, law enforcement personnel can find it virtually impossible to efficiently and completely search old case files each time they identify a new suspect. To this end, compiling DNA information into a searchable database gives law enforcement agencies a powerful tool to help identify, and potentially close, additional linked cases.
Processing and gaining meaningful insight from the staggering amount of data critical to homeland security has proven difficult. (10) Law enforcement agencies can use data mining to help them face this challenge.
For instance, investigators would like to anticipate, and thereby prevent, acts of terrorism. By using data mining to identify relevant historical patterns, trends, and relationships involving terrorists, they could accomplish this objective more effectively.
Also, because data mining allows law enforcement agencies to evaluate information in varied formats and from various databases and agencies, it can enable them to effectively and efficiently analyze a wide range of information that potentially could shed light on terrorist activity. For example, by analyzing information from multiple health-related data sources, law enforcement agencies could recognize significant patterns of illness that may indicate bioterrorism activity or the use of other weapons of mass destruction. (11) Agencies also can use this capability to associate general crimes with terrorist activity by linking them with additional intelligence--recent information suggesting links between cigarette smuggling and terrorist financing (12) serves as a valid example.
The law enforcement community may find that the capability of data mining in characterizing and monitoring normal activity, as well as identifying irregular or suspicious activity, proves applicable in the area of Internet and infrastructure protection. For example, the recognition of suspicious patterns of Web site activity not only can help in the area of traditional intrusion protection but also can serve as an important warning about the release of information. The FBI's National Infrastructure Protection Center (NIPC) recently underscored the importance of reviewing all Internet materials currently available, as well as those considered for release, for potential threats to critical infrastructure and homeland security. (13) This warning comes as many municipal Web sites are receiving suspicious activity and interest. (14) This information particularly includes that which, either on its own merits or in combination with other open-source materials, may prove useful to entities with malicious intent.
Law enforcement agencies face an ever-increasing flood of information that threatens to overwhelm them; this will require a change in how they process and analyze data. Data-mining technology represents a powerful, user-friendly, and accessible new tool that agencies can use to help them in facing this challenge as they seek to fulfill their missions--ultimately, to ensure the safety and welfare of the public.
(1) Tabassum Zakaria, "CIA Turns to Data Mining"; retrieved on April 10, 2003, from http://www.parallaxresearch.com/news/2001/0309/cia_turns_to.html.
(2) The authors based this article largely on their experience with and research on the subject of data mining.
(3) Bruce Moxon, "Defining Data Mining"; retrieved on April 10, 2003, from http://www.dbmsmag.com/9608d53.html.
(4) Law enforcement agencies must address appropriate constitutional and legal concerns if using public source data for law enforcement purposes.
(5) Law enforcement agencies, when collecting information from different sources, must decide how they will address the issue of cleaning the data, or preparing data for data-mining activities.
(6) Donald Brown, "The Regional Crime Analysis Program (RECAP): A Framework for Mining Data to Catch Criminals"; retrieved on April 10, 2003, from http://vijis.sys.virginia.edu/publication/RECAP.pdf.
(7) Ayse Imrohoroglu, Anthony Merlo, and Peter Rupert, "What Accounts for the Decline in Crime?"; retrieved on April 10, 2003, from http://www.clev.frb.org/Research/workpaper/2000/wp0008.pdf.
(8) Colleen McCue, Georgia Smith, Robyn Diehl, Deanne Dabbs, James McDonough, and Paul Ferrara, "Why DNA Databases Should Include All Felons," Police Chief, October 2001, 94-100.
(9) Richard Adderley and Peter Musgrove, "Data Mining Case Study: Modelling the Behaviour of Offenders Who Commit Serious Sexual Assaults," in Proceedings of the Seventh Association for Computing Machinery (ACM) Special Interest Group on Knowledge Discovery in Data (SIGKDD) International Conference on Knowledge Discovery and Data Mining Held in San Francisco 26-29 August 2001, (New York NY: ACM Press, 2001), 215-220.
(10) Supra note 1; and Eric Chabrow, "The FBI Must Overhaul Its IT Infrastructure to Fulfill a New Mandate of Fighting Terrorism, Cyberattacks"; retrieved on April 10, 2003, from http://www.informationweek.com/story/IWK20020602S0004; and Walter Pincus and Dana Priest, "NSA Intercepts on Eve of 9/11 Sent Warning"; retrieved on April 10, 2003, from http://www.secretpolicy.com/archives/00000073.html.
(11) Steve Bunk, "Early Warning: U.S. Scientists Counter Bioterrorism with New Electronic Sentinel Systems"; retrieved on April 10, 2003, from http://www.scenpro.com/press%2009%20leaders.html.
(12) Paul Nowell, "Hezbollah in North Carolina?"; retrieved on April 10, 2003, from http://www.abcnews.go.com/sections/us/dailynews/hezbollah010328.html.
(13) National Center for Infrastructure Protection (NIPC), Highlights, Issue 11-01, December 7, 2001; retrieved on April 10, 2003, from http://www.nipc.gov/publications/highlights/2001/highlight-01-11.htm.
(14) Barton Gellman, "Cyber-Attacks by Al Qaeda Feared"; retrieved on April 10, 2003, from http://www.washingtonpost.com/wp-dyn/articles/A50765-2002 Jun26.html.
By COLLEEN McCUE, Ph.D., EMILY S. STONE, M.S.W., and TERESA P. GOOCH, M.S.
Dr. McCue is the program manager for the Crime Analysis Unit of the Richmond, Virginia, Police Department and holds faculty appointments at Virginia Commonwealth University.
Lieutenant Colonel Gooch serves as assistant chief of the Richmond, Virginia, Police Department.
Ms. Stone served as a crime analyst with the Richmond, Virginia, Police Department.
|Printer friendly Cite/link Email Feedback|
|Author:||Gooch, Teresa P.|
|Publication:||The FBI Law Enforcement Bulletin|
|Date:||Nov 1, 2003|
|Previous Article:||The Bulletin Notes.|
|Next Article:||Patrol Operations and Enforcement Tactics.|