Classification System for Serial Criminal Patterns

Last updated

The Classification System for Serial Criminal Patterns (CSSCP) is an artificial intelligence computer system that assists law enforcement officials in identifying links between serial crimes. [1] Working in conjunction with a neural network called a Kohonen network, CSSCP finds patterns in law enforcement databases by analyzing the characteristics of an offender, the criminal activities that have occurred, and the objects used in a crime. Once the links between crimes have been identified by CSSCP, law enforcement officials can then use the data that is produced to build leads or solve criminal cases. Through its capability to run autonomously, the CSSCP has proven that it can operate non-stop without any human interaction and can achieve results with much more accuracy and efficiency than a human. [2]

Contents

Background

The Classification System for Serial Criminal Patterns was started by Professor Thomas Muscarello and Professor Kamal Dahbur at DePaul University in Chicago, Illinois in 1996 with the help of a Chicago Police Detective. Muscarello and Dahbur recognized the need for their hybrid system as a result of recognized deficiencies in police practices. These deficiencies were said to be a result of the police's difficulties in analyzing data and transforming it into information that could be useful in the investigation of crimes. [3] In addition to assisting law enforcement officials interpret data, the CSSCP was also designed to help investigators determine which criminal data was critical to an investigation and which format it should be stored in.

Although a similar project was previously undertaken by Timothy O'Shea, Muscarello and Dahbur noticed that problems existed with this proposed system because it relied on limited pre-processing and complex algorithms that lead to computational problems. [3]

The objective of Professor Muscarello and Professor Dahbur was to create an "automated methodology that can systematically identify groups of records as potential patterns for serial criminals, with a good degree of accuracy". [3]

System Design

The CSSCP program was designed to work in three separate phases in conjunction with a Kohonen network. The three phases of CSSCP include:

Diagram showing phases of CSSCP. CSSCP Diagram.svg
Diagram showing phases of CSSCP.
  1. Pre-processing phase
  2. Neural network phase
  3. Heuristics system phase

Pre-Processing Phase

The Pre-Processing Phase of CSSCP is considered to be the most important phase of the system because it is the phase in which data is structured and put in a format that can be used by the neural network. [3] This phase is where CSSCP will analyze the records it's provided, detect patterns among the data, and assign the data values according to the algorithm chosen. The pre-processing phase relies heavily on accurate and complete input data in order for the output data (results) to also be accurate and complete.

Within the pre-processing phase of CSSCP there are four major functions that take place in order to ensure that input data is processed correctly for the next phases. The four functions include:

  • Providing categories for missing values- In CSSCP, when a characteristic of a crime comes up that has not been previously assigned a value, the pre-processing phase will ensure that a category is created to account for that value so that it can be grouped with other characteristics of the same value in the future.
  • Standardization- By standardizing all the values (both discrete and continuous), the data can be better analyzed and a measure can be given to their significance in relation to other values. The standardization of all attributes serves the very important purpose of closing down the gaps among their respective domains. [3]
  • Categorizations & re-categorization- The Categorizing and re-categorizing of values and variables allows for users of CSSCP to determine a relationship between the values and variables and to create broader categories for data to be classified in. This function of CSSCP was created so that fewer categories could be used and so that the values could have more meaning to the user.
  • Grouping- Grouping in CSSCP is conducted so that discovering patterns can be done more effectively. Grouping allows adjustments to be made to the variables so that patterns can be depicted.

Neural Network Phase

The Neural Network phase of CSSCP is where machine learning algorithms are used for clustering and prediction tasks. [3] The type of neural network used in CSSCP is called a Kohonen network. The Kohonen network is an unsupervised learning network that uses a simple algorithm to make computations without needing to know the type or number of classifications to be used. [4]

Example of a Multi-Layer Neural Network Multi-Layer Neural Network-Vector.svg
Example of a Multi-Layer Neural Network

In the neural network phase, all the attributes and features of a case are grouped according to their given values (similarity) and each group is then presented to an independent Kohonen network. The independent Kohonen networks provide output independently and in parallel with the other independent networks in the system. [3] Once presented to the independent Kohonen networks, the groups are then combined for a final time and sent to a final Kohonen network. After being sent to the final Kohnen network, the system will suggest the preliminary classifications that will be sent on to the next and final phase. By the end of the neural network phase, all of the input data will have been analyzed, grouped, and classified into patterns that will become the basis for which the final results depend on.

Heuristics System Phase

The heuristics system phase is the final phase of the CSSCP application and is the stage at which the output data is refined. During this phase, the preliminary classifications that were developed in the previous phase are enhanced in order to improve classifications or to eliminate deficiencies. In CSSCP, heuristics are used for two main purposes:

  1. Adding records to a class
  2. Extracting records from a class

CSSCP uses heuristics to add records to a class that have been left out by the neural network. In CSSCP in particular, crimes that occur before midnight and after midnight would not be grouped together during the neural network phase regardless of how many similarities exists between them. The use of heuristics corrects this problem by adding an appropriate group for both crimes to be grouped together. [3]

CSSCP also uses heuristics to extract or remove records from a class in two ways. One way is in instances where the neural network has grouped two or more specific crimes together that cannot be related to each other. The second way is in instances when classifications have been created that are irrelevant to the user; such as when a group is created to compare characteristics of multiple criminals involved in the same crime as opposed to a group that compares the characteristics of the crime itself.

Applications

The first use of CSSCP was during a trial study in which statistics from three years of armed robbery cases were analyzed. Professor Muscarello decided to demonstrate with armed robbery cases not only because they were some of the most frequent cases dealt with by police, but also because they tend to demonstrate common patterns that would highlight the abilities of CSSCP. [3] During the trial, CSSCP demonstrated how it could correctly classify patterns at a much higher rate than the suggested one percent that a Rand Corporation study reported as the common rate. [5] Because of CSSCP's success during the trial studies, the Chicago Police Department decided to conduct live trials of CSSCP on their networks in 2006. However, due to upper management changes within the Chicago Police Department and the retirement of the Deputy Superintendent who sponsored the project, CSSCP was never actually put into use by the department.[ citation needed ]

The CSSCP program is designed to assist law enforcement officials who constantly deal with large volumes of criminal cases beyond what their departments can effectively handle by providing them with an inexpensive tool that can reduce investigation costs and department man-power. [5] However, the application of CSSCP does not intend to entirely replace human detectives with a computer program, but instead intends to assist detectives by making their jobs easier and their workloads lighter. [6]

Through its ability to continuously operate accurately at a rate of ten times faster than a team of detectives doing the same type of work, CSSCP has begun to draw interest from law enforcement agencies all over the world that are looking for tools that can enhance security. [6]

Related Research Articles

<span class="mw-page-title-main">Neural network (machine learning)</span> Computational model used in machine learning, based on connected, hierarchical functions

In machine learning, a neural network is a model inspired by the structure and function of biological neural networks in animal brains.

<span class="mw-page-title-main">Self-organizing map</span> Machine learning technique useful for dimensionality reduction

A self-organizing map (SOM) or self-organizing feature map (SOFM) is an unsupervised machine learning technique used to produce a low-dimensional representation of a higher-dimensional data set while preserving the topological structure of the data. For example, a data set with variables measured in observations could be represented as clusters of observations with similar values for the variables. These clusters then could be visualized as a two-dimensional "map" such that observations in proximal clusters have more similar values than observations in distal clusters. This can make high-dimensional data easier to visualize and analyze.

Pattern recognition is the task of assigning a class to an observation based on patterns extracted from data. While similar, pattern recognition (PR) is not to be confused with pattern machines (PM) which may possess (PR) capabilities but their primary function is to distinguish and create emergent patterns. PR has applications in statistical data analysis, signal processing, image analysis, information retrieval, bioinformatics, data compression, computer graphics and machine learning. Pattern recognition has its origins in statistics and engineering; some modern approaches to pattern recognition include the use of machine learning, due to the increased availability of big data and a new abundance of processing power.

Machine learning (ML) is a field of study in artificial intelligence concerned with the development and study of statistical algorithms that can learn from data and generalize to unseen data and thus perform tasks without explicit instructions. Recently, artificial neural networks have been able to surpass many previous approaches in performance.

Unsupervised learning is a framework in machine learning where, in contrast to supervised learning, algorithms learn patterns exclusively from unlabeled data. Other frameworks in the spectrum of supervisions include weak- or semi-supervision, where a small portion of the data is tagged, and self-supervision. Some researchers consider self-supervised learning a form of unsupervised learning.

In computer science, learning vector quantization (LVQ) is a prototype-based supervised classification algorithm. LVQ is the supervised counterpart of vector quantization systems.

Geographic profiling is a criminal investigative methodology that analyzes the locations of a connected series of crimes to determine the most probable area of offender residence. By incorporating both qualitative and quantitative methods, it assists in understanding spatial behaviour of an offender and focusing the investigation to a smaller area of the community. Typically used in cases of serial murder or rape, the technique helps police detectives prioritize information in large-scale major crime investigations that often involve hundreds or thousands of suspects and tips.

<span class="mw-page-title-main">Crime mapping</span> Diagram showing crime incident patterns

Crime mapping is used by analysts in law enforcement agencies to map, visualize, and analyze crime incident patterns. It is a key component of crime analysis and the CompStat policing strategy. Mapping crime, using Geographic Information Systems (GIS), allows crime analysts to identify crime hot spots, along with other trends and patterns.

When classification is performed by a computer, statistical methods are normally used to develop the algorithm.

Criminal intelligence is information compiled, analyzed, and/or disseminated in an effort to anticipate, prevent, or monitor criminal activity.

<span class="mw-page-title-main">Crime analysis</span>

Crime analysis is a law enforcement function that involves systematic analysis for identifying and analyzing patterns and trends in crime and disorder. Information on patterns can help law enforcement agencies deploy resources in a more effective manner, and assist detectives in identifying and apprehending suspects. Crime analysis also plays a role in devising solutions to crime problems, and formulating crime prevention strategies. Quantitative social science data analysis methods are part of the crime analysis process, though qualitative methods such as examining police report narratives also play a role.

Product finders are information systems that help consumers to identify products within a large palette of similar alternative products. Product finders differ in complexity, the more complex among them being a special case of decision support systems. Conventional decision support systems, however, aim at specialized user groups, e.g. marketing managers, whereas product finders focus on consumers.

Crime prevention is the attempt to reduce and deter crime and criminals. It is applied specifically to efforts made by governments to reduce crime, enforce the law, and maintain criminal justice.

In network theory, link analysis is a data-analysis technique used to evaluate relationships between nodes. Relationships may be identified among various types of nodes (100k), including organizations, people and transactions. Link analysis has been used for investigation of criminal activity, computer security analysis, search engine optimization, market research, medical research, and art.

Multimedia information retrieval is a research discipline of computer science that aims at extracting semantic information from multimedia data sources. Data sources include directly perceivable media such as audio, image and video, indirectly perceivable sources such as text, semantic descriptions, biosignals as well as not perceivable sources such as bioinformation, stock prices, etc. The methodology of MMIR can be organized in three groups:

  1. Methods for the summarization of media content. The result of feature extraction is a description.
  2. Methods for the filtering of media descriptions
  3. Methods for the categorization of media descriptions into classes.

The fields of marketing and artificial intelligence converge in systems which assist in areas such as market forecasting, and automation of processes and decision making, along with increased efficiency of tasks which would usually be performed by humans. The science behind these systems can be explained through neural networks and expert systems, computer programs that process input and provide valuable output for marketers.

In the United States, the practice of predictive policing has been implemented by police departments in several states such as California, Washington, South Carolina, Alabama, Arizona, Tennessee, New York, and Illinois. Predictive policing refers to the usage of mathematical, predictive analytics, and other analytical techniques in law enforcement to identify potential criminal activity. Predictive policing methods fall into four general categories: methods for predicting crimes, methods for predicting offenders, methods for predicting perpetrators' identities, and methods for predicting victims of crime.

<span class="mw-page-title-main">Spatiotemporal pattern</span> Patterns in both time and space

Spatiotemporal patterns are patterns that occur in a wide range of natural phenoma and are characterized by a spatial and temporal patterning. The general rules of pattern formation hold. In contrast to "static", pure spatial patterns, the full complexity of spatiotemporal patterns can only be recognized over time. Any kind of traveling wave is a good example of a spatiotemporal pattern. Besides the shape and amplitude of the wave, its time-varying position in space is an essential part of the entire pattern.

This glossary of artificial intelligence is a list of definitions of terms and concepts relevant to the study of artificial intelligence, its sub-disciplines, and related fields. Related glossaries include Glossary of computer science, Glossary of robotics, and Glossary of machine vision.

<span class="mw-page-title-main">Predictive policing</span> Use of predictive analytics to direct policing

Predictive policing is the usage of mathematics, predictive analytics, and other analytical techniques in law enforcement to identify potential criminal activity. A report published by the RAND Corporation identified four general categories predictive policing methods fall into: methods for predicting crimes, methods for predicting offenders, methods for predicting perpetrators' identities, and methods for predicting victims of crime.

References

  1. ↑ The Crime Report, (2005). "Super Computer Program To Track Chicago Crime Patterns". www.thecrimereport.org. Retrieved on 2011-10-16. Available @ http://www.thecrimereport.org/archive/super-computer-program-to-track-chicago-crime-patterns.
  2. ↑ Patrice M. Jones, (2005). "The cyber sleuth". Chicago Tribune. Retrieved on 2011-10-17. Available @ .
  3. 1 2 3 4 5 6 7 8 9 Kamal Dahbur and Thomas Muscarello, (2004). "Classification system for serial criminal patterns". Artificial Intelligence and Law 11: 251–269.
  4. ↑ Kohonen, T. (1990). The Self-Organizing Map, Proceedings of IEEE, 78 (9).
  5. 1 2 DePaul University, (2004). "DePaul Researchers Develop Computer Program to Assist Police in Identifying Patterns of Criminal Activity". DePaul University. Retrieved on 2011-10-18. Available @ https://newsroom.depaul.edu/NewsReleases/showNewsPrint.aspx?NID=1258.
  6. 1 2 The Age, (2005). "Tech comes to the aid of crimefighters". www.theage.com.au. Retrieved on 2011-10-22. Available @ http://www.theage.com.au/news/technology/tech-comes-to-the-aid-of-crimefighters/2005/12/12/1134235994179.html.
This page is based on this Wikipedia article
Text is available under the CC BY-SA 4.0 license; additional terms may apply.
Images, videos and audio are available under their respective licenses.