Homa Atabakhsh

COPLINK Connect: information and knowledge management for law enforcement

As part of nationwide, ongoing digital government initiatives, COPLINK [Chen et al. 2002, Chen et al. 2003, Hauck et al. 2002] is an integrated information and knowledge management environment aimed at meeting some of the challenges faced by the law enforcement community. Funded by the National Institute of Justice and the National Science Foundation, a prototype for COPLINK was initially developed at the University of Arizonas Artificial Intelligence Lab in collaboration with the Tucson Police Department (TPD) and Phoenix Police Department (PPD). COPLINK was developed into a product by Knowledge Computing Corporation (KCC) and deployed in approximately one hundred law enforcement agencies nationwide [see for example The Los Angeles Daily News Dec. 6, 2003 and Anchorage Daily News Nov. 23, 2003].

Automatically detecting deceptive criminal identities

The criminal mind is no match for some of the latest technology designed to determine fact from fiction in suspect identities.

Information sharing and collaboration policies within government agencies

This paper describes the necessity for government agencies to share data as well as obstacles to overcome in order to achieve information sharing. We study two domains: law enforcement and disease informatics. Some of the ways in which we were able to overcome the obstacles, such as data security and privacy issues, are explained. We conclude by highlighting the lessons learned while working towards our goals.

Discovering identity problems: a case study

Identity resolution is central to fighting against crime and terrorist activities in various ways. Current information systems and technologies deployed in law enforcement agencies are neither adequate nor effective for identity resolution. In this research we conducted a case study in a local police department on problems that produce difficulties in retrieving identity information. We found that more than half (55.5%) of the suspects had either a deceptive or an erroneous counterpart existing in the police system. About 30% of the suspects had used a false identity (i.e., intentional deception), while 42% had records alike due to various types of unintentional errors. We built a taxonomy of identity problems based on our findings.

Visualization in law enforcement

Visualization techniques have proven to be critical in helping crime analysis. By interviewing and observing Criminal Intelligence Officers (CIO) and civilian crime analysts at the Tucson Police Department (TPD), we found that two types of tasks are important for crime analysis: crime pattern recognition and criminal association discovery. We developed two separate systems that provide automatic visual assistance on these tasks. To help identify crime patterns, a Spatial Temporal Visualization (STV) system was designed to integrate a synchronized view of three types of visualization techniques: a GIS view, a timeline view and a periodic pattern view. The Criminal Activities Network (CAN) system extracts, visualizes and analyzes criminal relationships using spring-embedded and blockmodeling algorithms. This paper discusses the design and functionality of these two systems and the lessons learned from the development process and interaction with law enforcement officers.

A multi-layer Naïve bayes model for approximate identity matching

Identity management is critical to various governmental practices ranging from providing citizens services to enforcing homeland security. The task of searching for a specific identity is difficult because multiple identity representations may exist due to issues related to unintentional errors and intentional deception. We propose a Naive Bayes identity matching model that improves existing techniques in terms of effectiveness. Experiments show that our proposed model performs significantly better than the exact-match based technique and achieves higher precision than the record comparison technique. In addition, our model greatly reduces the efforts of manually labeling training instances by employing a semi-supervised learning approach. This training method outperforms both fully supervised and unsupervised learning. With a training dataset that only contains 30% labeled instances, our model achieves a performance comparable to that of a fully supervised learning.

A hierarchical Naïve Bayes model for approximate identity matching

Organizations often manage identity information for their customers, vendors, and employees. Identity management is critical to various organizational practices ranging from customer relationship management to crime investigation. The task of searching for a specific identity is difficult because disparate identity information may exist due to the issues related to unintentional errors and intentional deception. In this paper we propose a hierarchical Naive Bayes model that improves existing identity matching techniques in terms of searching effectiveness. Experiments show that our proposed model performs significantly better than the exact-match based matching technique. With 50% training instances labeled, the proposed semi-supervised learning achieves a performance comparable to the fully supervised record comparison algorithm. The semi-supervised learning greatly reduces the efforts of manually labeling training instances without significant performance degradation.

COPLINK: Information and knowledge management for law enforcement

The problem of information and knowledge management in the knowledge intensive and time critical environment of law enforcement has posed an interesting problem for information technology professionals in the field. Coupled with this challenging environment are issues relating to the integration of multiple systems, each having different functionalities resulting in difficulty for the end user. COPLINK offers a cost-efficient way of web enabling stovepipe law enforcement information sharing systems by employing a model for allowing different police departments to more easily share data amongst themselves through an easy-to-use interface that integrates different data sources. The COPLINK project has two major components: COPLINK Database Application and COPLINK Concept Space Application.

COPLINK center: social network analysis and identity deception detection for law enforcement and homeland security intelligence and security informatics: a crime data mining approach to developing border safe research

In this paper, we describe the highlights of the COPLINK Center for law enforcement and homeland security project. Two new components of the project are described, namely, identity resolution and mutual information.

A probabilistic model for approximate identity matching

Identity management is critical to various governmental practices ranging from providing citizens services to enforcing homeland security. The task of searching for a specific identity is difficult because multiple identity representations may exist due to issues related to unintentional errors and intentional deception. We propose a probabilistic Naïve Bayes model that improves existing identity matching techniques in terms of effectiveness. Experiments show that our proposed model performs significantly better than the exact-match based technique as well as the approximate-match based record comparison algorithm. In addition, our model greatly reduces the efforts of manually labeling training instances by employing a semi-supervised learning approach. This training method outperforms both fully supervised and unsupervised learning. With a training dataset that only contains 10% labeled instances, our model achieves a performance comparable to that of a fully supervised learning.