没有合适的资源?快使用搜索试试~ 我知道了~
A Methodology for Evaluating Geographic Profiling Software
需积分: 0 1 下载量 60 浏览量
2010-02-21
14:52:54
上传
评论
收藏 2.12MB PDF 举报
温馨提示
试读
259页
A Methodology for Evaluating Geographic Profiling Software
资源详情
资源评论
资源推荐
The author(s) shown below used Federal funds provided by the U.S.
Department of Justice and prepared the following final report:
Document Title: A Methodology for Evaluating Geographic
Profiling Software
Author(s): Tom Rich ; Michael Shively
Document No.: 208993
Date Received: December 2004
Award Number: ASP T-037
This report has not been published by the U.S. Department of Justice.
To provide better customer service, NCJRS has made this Federally-
funded grant final report available electronically in addition to
traditional paper copies.
Opinions or points of view expressed are those
of the author(s) and do not necessarily reflect
the official position or policies of the U.S.
Department of Justice.
A Methodology for
Evaluating Geographic
Profiling Software
Cambridge, MA
Lexington, MA
Hadley, MA
Bethesda, MD
Chicago, IL
Draft Final Report
November 2004
Prepared for
Ronald Wilson
National Institute of Justice
810 Seventh Street, N.W.
Washington, DC 20531
Prepared by
Tom Rich
Abt Associates Inc.
55 Wheeler Street
Cambridge, MA 02138
Michael Shively, Ph.D.
This document is a research report submitted to the U.S. Department of Justice. This report has not
been published by the Department. Opinions or points of view expressed are those of the author(s)
and do not necessarily reflect the official position or policies of the U.S. Department of Justice.
Internal Review
Project Director
Technical Reviewer
Management Reviewer
This document is a research report submitted to the U.S. Department of Justice. This report has not
been published by the Department. Opinions or points of view expressed are those of the author(s)
and do not necessarily reflect the official position or policies of the U.S. Department of Justice.
Contents
1.
Introduction................................................................................................................................1
1.1. Background on Geographic Profiling................................................................................1
1.2. Approach to Methodology Development ..........................................................................3
2. Expert Panel Summary .............................................................................................................5
2.1. Day 1 .................................................................................................................................5
2.2. Day 2 .................................................................................................................................8
3. Background on the Geographic Profiling Software Applications.......................................11
3.1. CrimeStat.........................................................................................................................12
3.2. Dragnet ............................................................................................................................13
3.3. Predator............................................................................................................................13
3.4. Rigel Analyst...................................................................................................................14
4. Evaluation Methodology .........................................................................................................15
4.1. Summary..........................................................................................................................15
4.2. Output Accuracy Testing.................................................................................................16
Test Inputs .......................................................................................................................16
Performance Measures ....................................................................................................17
Data Analysis...................................................................................................................18
4.3. User Feedback .................................................................................................................18
User Survey .....................................................................................................................18
User Log..........................................................................................................................19
4.4. Feature Analysis..............................................................................................................20
4.5. Summary of Deliverables................................................................................................21
References ...........................................................................................................................................22
Appendix: Expert Panel Transcript .................................................................................................27
Abt Associates Inc. Contents i
This document is a research report submitted to the U.S. Department of Justice. This report has not
been published by the Department. Opinions or points of view expressed are those of the author(s)
and do not necessarily reflect the official position or policies of the U.S. Department of Justice.
1. Introduction
This report describes a methodology for evaluating geographic profiling software. Following a brief
overview of geographic profiling (Section 1.1), Section 1.2 describes how the methodology was
developed. The key component of the methodology was convening an expert panel that met in
August 2004; a summary and full transcript of the panel’s discussions are in Section 2 and the
Appendix, respectively. The panel focused on four geographic profiling software applications, which
are described in Section 3. The actual evaluation methodology is outlined in Section 4.
1.1. Background on Geographic Profiling
Geographic profiling is a criminal investigative technique that attempts to provide information on the
likely “base of operations” of offenders thought to be committing serial crimes. The base of
operations could be the offender’s home, place of employment, a friend house, or some other
frequented location. The predictions are based on the locations of these crimes, other geographic
information about the case and the suspect, and certain assumptions about the distance offenders will
travel to commit crimes.
Canter (2003) argues that geographical profiling was “born” in 1980 when a UK police investigator
analyzed the locations of crime scenes of the Yorkshire Ripper and computed the “center of gravity”
of the crime scenes thought to be linked to the case. It turned out that the offender lived in the town
that the investigator predicted. No doubt other investigators and crime analysts have approximated
such information by visual inspection ever since the advent of paper “pin maps.”
In the mid-1990s, more sophisticated models for predicting an offender’s home address were
developed, building on the work of Brantingham and Brantingham (1981) and other studies of
offender travel behavior (e.g., Rhodes and Conly, 1981). As summarized in Rossmo (1999), key
results of these studies include:
• Most crimes occur in relatively close proximity to the offender’s home.
• Crime trips follow a distance-decay function, with the number of crime occurrences
decreasing with distance from the offender’s home.
• Juvenile offenders exhibit less mobility than adult offenders
• Patterns in crime trip distances vary by crime type.
Rossmo (1995, 1998, 1999), in particular, extended the work of the Brantinghams and developed a
“criminal geographic targeting” algorithm, which was later patented and incorporated into the Rigel
software application. Levine (2002, p. 357) indicates that the journey-to-crime routines in CrimeStat
“builds on the Rossmo framework, but extends its modeling capability.” Canter (1999, 2004)
developed his Dragnet software in the mid-1990s based on his work with police investigators in the
UK.
While Rigel, CrimeStat, and Dragnet are based on different types of distance-decay functions, they
produce the same general type of output. In contrast to a single spatial mean (used in the Yorkshire
Ripper case), these software applications create a grid over an area and then calculate the probability
that the offender’s base of operations is in each grid cell based on the specified crime-related
Abt Associates Inc. Draft Evaluation Methodology for Geographic Profiling Software 1
This document is a research report submitted to the U.S. Department of Justice. This report has not
been published by the Department. Opinions or points of view expressed are those of the author(s)
and do not necessarily reflect the official position or policies of the U.S. Department of Justice.
剩余258页未读,继续阅读
sea_star666
- 粉丝: 0
- 资源: 4
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功
评论0