Gumshoe: a model for undergraduate computational ...

Download PDF
1 downloads 0 Views 250KB Size Report
Comment
for illegal gun possession in the city never faced trial. Philadelphia''s ..... firearm that, under Pennsylvania law, theoretically should result in a convicted gun ...
Gumshoe: A Model for Undergraduate Computational Journalism Education Sarah Monisha Pulimood

Donna Shaw

Emilie Lounsberry

Dept. of Computer Science The College of New Jersey Ewing, NJ 08628 +1 609 771 2788

Dept. of English The College of New Jersey Ewing, NJ 08628 +1 609 771 2793

Dept. of English The College of New Jersey Ewing, NJ 08628 +1 609 771 2506

[email protected]

[email protected]

[email protected]

immediate outcry from readers and politicians alike, with a number of city, state and federal officials stepping in to investigate and promising to find remedies. Those efforts, including interventions by the Pennsylvania Supreme Court and U.S. Senator Arlen Specter, are ongoing as we write this paper. The series also has received recognition by state and national press organizations. What readers of The Inquirer do not know is that the series has its roots in a computer assisted reporting (CAR) project, Gumshoe, conducted in 2007 by undergraduate computer science and journalism students and their professors at our small, primarily undergraduate liberal arts college. In this paper, we describe the collaboration between students and faculty in our computer science and journalism programs, and the large metropolitan newspaper that found in our students’’ work a starting point for what ultimately became a hard-hitting series of investigative stories, with far-reaching consequences.

ABSTRACT This paper describes a collaboration between computer science and journalism students and professors at our small, primarily undergraduate college, and a large metropolitan newspaper. Our students’’ work was a catalyst for a hard-hitting series of investigative stories, with far-reaching consequences. The Gumshoe project is a model for computational journalism at an undergraduate institution. The project demonstrates that when computer scientists and journalists reach out across disciplinary boundaries, computational thinking and collaboration can solve a real problem, and have a substantive impact on society.

Categories and Subject Descriptors K.3.2 [Computing Milieux] Computer and Information Science Education K. 4.0 Computers and Society

2. BACKGROUND Computer science educators recognize that rapid advances in technology and paradigm shifts in the global economy are giving rise to new models for how information is gathered, disseminated and used, and that it is essential to expose students from all disciplines to the ideas of computational thinking [1, 18]. While there is as yet no uniform definition, there is general agreement that computational thinking includes a broad range of mental tools and concepts from computer science that help people solve problems, design systems, understand human behavior, and engage computers to assist in automating a wide range of intellectual processes [1]. A recent report from the Committee for the Workshops on Computational Thinking convened by the National Research Council at the National Academies describes computational thinking as a group phenomenon as well as an individual one and this capability can assist specialists in other disciplines to more effectively adopt, use, and develop computational tools [1]. In a world now dominated by Google, Facebook and Twitter, it is imperative that all students, not just computer science majors, understand how information propagation is influenced by databases. Almost every facet of our lives is touched by data that we must analyze in some form to make informed decisions. The natural implication of this trend is the increase in dependence on the credibility and validity of electronic information. As complexity of data increases, it is no longer sufficient to store the data in flat file structures, like spreadsheets; we must be able to retrieve data in interesting ways conducive to deeper analysis. Computing technologies such as databases, geographic information systems, statistical analysis programs and the Internet

General Terms Human Factors

Keywords Computer science education, computational thinking, journalism education, computer aided reporting (CAR), database systems, interdisciplinary collaboration.

1. INTRODUCTION For crime-weary citizens of Philadelphia, it may have come as little surprise when, in December 2009, the Philadelphia Inquirer published a four-part series, ““Justice: Delayed, Dismissed, Denied,”” which concluded that most suspects arrested for illegal gun possession in the city never faced trial. Philadelphia’’s criminal justice system, the newspaper concluded, was seriously out of control, warped by faulty prosecutions, inept court procedures, and savvy defense lawyers and defendants who had learned to ““play”” the system [11]. As detailed by the newspaper, the enormous number of mishandled cases and the flimsy reasons for their dismissals by city judges led to loud and

Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. SIGCSE’’11, March 9––12, 2011, Dallas, Texas, USA. Copyright 2011 ACM 978-1-4503-0500-6/11/03...$10.00.

529

journalism is increasingly dependent on ““reliable, downloadable, and searchable databases”” [3]. The report suggests that universities should develop ““innovations in news reporting and dissemination for the digital era,”” and since they are among the nation’’s largest nonprofit institutions, universities ““should play significant roles in the reconstruction of American journalism.”” Computational journalism is emerging as a discipline in its own right, gaining traction in higher education. For example, Georgia Tech2 offers a course on computational journalism and Columbia University recently announced a new dual Master of Science program in Computer Science and Journalism3. At Duke University, researchers and professors at the DeWitt Wallace Center for Media & Democracy are examining computational journalism, and journalists and computer scientists are working together to create software tools intended to help reporters move more quickly to produce in-depth, authoritative stories4. The Medill School of Journalism at Northwestern University is offering scholarships to lure programmers and web developers to their journalism master’’s degree program5. Models similar to our collaboration are being used elsewhere: Journalism students from the Medill Innocence Project at Northwestern University have helped uncover evidence that so far has freed 11 men from prison6. Former ‘‘60 Minutes’’ investigative producer Lowell Bergman, now a professor at the University of California at Berkeley Graduate School of Journalism, has forged an alliance between the New York Times and the PBS documentary program, ““Frontline,”” using his graduate students to assist in stories on corruption in Mexico, the East Africa bombings and the California energy crisis and the role of Enron7. The Los Angeles Times and the University of Southern California’’s (USC) Annenberg School for Communication & Journalism are collaborating to produce stories for the Times’’ Homicide Report blog. Under this partnership, USC students write dispatches for the Homicide Report, providing more content for the blog and more crime-reporting experience for students8. The New York Times announced that New York University students in a new class, ““The Hyperlocal Newsroom,”” would staff the ““East Village Local,”” starting in fall 20109, and in a partnership with the Boston Globe, students from Northeastern University have written 11 Page One stories in 20 months10. The Gumshoe project described here is a perfect example of how computer scientists and journalists, even at an undergraduate institution, can use computational thinking to collaborate and solve a real problem making a real impact on society.

have radically changed the way journalists investigate and deliver news to the public. In its 2010 report on the state of the news media, the Pew Project for Excellence in Journalism estimated that 5,900 full-time newsroom jobs were lost in 2009, in addition to a similar number in 2008 with resulting in the loss of almost a third of the workforce in American newspapers since 2001 [10]. Since 2001, newsrooms across the United States have lost more than 25 percent of staffers, bringing the total number of reporters, editors and other staffers employed in daily journalism down to 41,500 –– the lowest number since the 1970s, according to the American Society of Newspaper Editors1. While much of the staff reductions can be attributed to the dim economy, the drop is also due to lost revenue stemming from the increase in the public’’s shift to free online news. This underscores the reality that 21st century journalists, like many other professionals in computingdependent fields, increasingly require a strong computing background [5, 8, 9]. Journalists are departing from traditional story-centric journalism and moving toward the use of information that is structured and compiled by computers to enable more comprehensive and analytical investigations [5], improving the efficiency, accuracy, structuring and quality of the targeted journalistic endeavor over traditional methods and tools. Both computer scientists and journalists must be able to apply foundational principles of information access and dissemination, fact analysis, process description and decisionmaking for developing software artifacts and news items respectively. Although there is still some disagreement about how these goals can be accomplished, there is growing recognition among industry leaders that journalists must become technologically literate, and that technological innovation is essential to the fulfillment of the ethical and economic imperatives of the industry [8, 9]. Computer assisted reporting (CAR) has been used in newsrooms for decades. However, ““computational journalism”” requires a more sophisticated approach to applying algorithms and principles from computer science and the social sciences to gather, evaluate, organize and present news and information [6]. Reporters and editors need quicker access to information that will help them provide more in-depth reports, complete with history, context and the most recent supporting data. An example is the investigation performed by New York Times reporter Jo Craven McGinty. Using a database obtained from the New York Police Department Hate Crimes Task Force, McGinty comprehensively mapped numerous hate crimes in New York City based on type, location, frequency and targeted ethnicity. McGinty concluded that Jews are the targets of most NYC hate crimes and that most of them occur in Brooklyn [7]. Economic reality is forcing colleges and universities to reexamine how to educate young journalists, how to help find a sustainable business model for online news and to figure out what computer skills will be most helpful in newsgathering. Many journalists and journalism professors who once eschewed the idea that reporters and editors would need computer programming skills now realize that such skills are important, perhaps even essential, to the future of professional newsgathering. As former Washington Post top editor Leonard Downie Jr. and Columbia journalism professor Michael Schudson concluded in their report, ‘‘The Reconstruction of American Journalism’’, sophisticated 1

2

http://www.computation-and-journalism.com/main/ http://www.wired.com/epicenter/2010/04/will-columbia-trainedcode-savvy-journalists-bridge-the-mediatech-divide/ 4 http://dewitt.sanford.duke.edu/index.php/about/area-ofresearch/computational_journalism 5 http://www.medill.northwestern.edu/admissions/ page.aspx?id=58645 . 6 http://www.medillinnocenceproject.org/ 7 http://journalism.berkeley.edu/faculty/bergman/ 8 http://annenberg.usc.edu/News%20and%20Events/News/ 100127LATimesPartnership.aspx 9 http://phx.corporate-ir.net/phoenix.zhtml?c=105317&p=irolpressArticle&ID=1393695&highlight 10 http://www.nieman.harvard.edu/reportsitem.aspx?id=100934 3

http://asne.org/annual_conference/conference_news/articleid/ 763/decline-in-newsroom-jobs-slows-763.aspx

530

x x

3. THE COLLABORATION Our role in the gun crime project began in early 2007 as a discussion among colleagues at our institution. One of the coauthors is a computer science professor who teaches a course on database systems. An important goal of this course is for students to collaborate with other students to solve problems by integrating concepts from computer science, particularly databases, and other disciplines. Students are required to work in teams to design and implement a database system to address a real-world need in order to gain a deeper understanding of and internalize database modeling concepts such as data modeling, queries, normalization, transaction and concurrency control, as well as issues of privacy and security of data. By working with people or ““clients”” from other disciplines, students also better understand the techniques and process for effectively eliciting requirements for a system outside their familiar domain [14]. Another co-author is a journalism professor who, through an independent study course, sought to provide journalism students with the opportunity for and experience of a painstaking computer assisted reporting (CAR) project that is essential yet frequently too time-consuming and expensive for a newspaper, given the diminishing newsroom resources. In light of market realities, we have sought more opportunities to pair professional journalists with our students not only to train the next generation, but also to fill the investigative reporting gaps created by buyouts and layoffs in our region’’s print media. What followed was a semester-long collaboration between the nine computer science students in the database systems class and the nine journalism students in the independent study group. We chose to address violent gun crimes because they were prominent in the news and involved one of our areas of journalistic expertise. For the journalism independent study group the project was the sole deliverable for the course. The students worked on the project in class for at least three hours per week throughout the semester. Initially, an Excel spreadsheet was created that listed demographic data for each person arrested for violating 18 Pa. C.S.A. § 6106, carrying a firearm without a license, in a sample period of January-February 2006. That violation was chosen because almost everyone accused of a gun-related crime would be charged with at least that count. The only exceptions would be a handful of people with licenses to carry weapons. That resulted in a list of 699 people. Of that number, 613 defendants were tracked, since the rest involved juveniles whose cases went almost immediately to Family Court, or defendants whose records were expunged, or those whose charges were not in the defined timeframe. Using Pennsylvania’’s online court dockets, the journalism students painstakingly created and populated several more columns of data, for example felony charges like murder, manslaughter, aggravated assault, rape, etc. while in possession of a firearm that, under Pennsylvania law, theoretically should result in a convicted gun offender receiving a mandatory minimum prison sentence of five years. We quickly discovered that determining who should receive a mandatory sentence would be almost impossible, in large part because the dockets did not list the degree of the charges, even though there was space provided on the forms for that information. Instead, we tracked each case from the initial arrest, through preliminary hearing, to ultimate disposition (trial and sentencing, if any), to identify where the system was failing. Some questions we wanted to ask were: x What happens to defendants arrested for gun-related crimes? x How many people serve prison time for committing violent felonies with guns?

Are cases regularly being dismissed or bargained down? In how many cases do civilian witnesses and/or victims fail to appear in court to testify?

However, as the data grew, effective data management and analysis became more challenging. Excel, like other file-based approaches, has a number of limitations such as isolation of data, duplication of data, incompatible file formats, and fixed query proliferation [2, 4]. It was clear that a well-designed database system would go a long way in helping to analyze the data and find meaningful patterns and information. Moreover, it would eliminate data redundancy, improve data integrity, improve data accessibility, and enforce data consistency. The students in the Database Systems class were divided into two teams, and each was charged with the task of designing a version of the system, in a series of pre-defined stages. The goal of the project was defined as developing a CAR system to store, manage, track, and query data on gun crime in the Philadelphia region. Throughout the semester, as concepts were covered in class, they were applied to the next stage of the project. The project contributed to 25% of the final grade. However, as we discuss below, exploration and internalization of important concepts contributed indirectly to other aspects such as the assignment grade (20%) and overall understanding of material. Periodically, representatives from each computer science team met with the journalism team to clarify their own understanding and to confirm that their design was progressing in a way that continued to meet the journalistic needs. This was particularly essential as they designed the queries that would retrieve in the data in ways that would be useful and enlightening for the journalists. The computer science professor closely monitored and guided the progress to ensure that pedagogical and system goals were being met. Both versions of the project were completed successfully and presented to the journalism professor. Feedback from the journalism professor indicated that there were useful and compelling features in each of the projects. Elements from each version were integrated into a final system, Gumshoe, in the following semester by one of the computer science students who undertook the task as part of a mentored research project under the guidance of the computer science professor.

4. SYSTEM DESIGN 4.1 Data querying and organization A number of data querying and organization methodologies were analyzed to determine which would be most effective for the CAR system. One methodology entails organizing data by the specific entity types pertinent to the investigation and tailoring search and sort algorithms to only these entities. This gives the journalist the ability to analyze all the attributes and information about a category of objects within the system, thus enabling focused analysis of a specific entity [2, 4]. Various entity categories concerning gun crimes and criminals such as arrest, defendants, and courts were identified and algorithms that search and organize all objects by the identified entity types were designed. Another methodology entails implementing the ability to search for a specific entity either by value or attribute within the collective data set [2, 4]. This is a contrast to the macro level data organization of the previous method because it focuses on a specific object and the attributes associated with that object. This is a desirable feature for CAR because it allows the journalist to search for a very specific entity or criteria on a micro level. It is also a powerful tool that enables the journalist to narrow down the

531

analysis and precisely examine a more specific investigative query rather than general patterns. An example would be searching for an identification number of a specific criminal to view his or her record. A third methodology enables searching by the characteristics of an object type and comparing the results to other object types. Thus, the search attributes of one object can be compared to the search attributes of a different object in the system. The searching of characteristics or attributes of an object is a middle ground between macro level and micro level searches because it is based on more criteria shared by objects but does not analyze a single record. Most important, this methodology represents a fundamental analytical process common to many journalistic investigations: drawing conclusions by comparing events to similar cases. As an example, one could compare the verdicts decided by one judge to the verdicts of another judge. One technique is to design the database schema or underlying platform of the application to be data neutral with a generic representation of the data to be analyzed. This is a powerful concept because it enables the importing, expansion, and analysis of other data sets in the system for subsequent investigations. Yet another method is to allow data in the system to be exported outside the system boundary. This is important because it gives the journalist greater investigative flexibility by allowing for analysis in other software. For example, Gumshoe allows data to be exported to files that can be opened in Microsoft Excel for graph generation and statistical analysis, abilities not native to Gumshoe’’s functional scope.

As the class considered the data to be stored, such as demographics for the User entity, important concepts of security and issues of privacy of data were examined in class. For example, we examined the necessity and appropriateness of storing the social security number for a defendant, and if we did, which types of users should be allowed to access it. This raised interesting questions about the legal rights of defendants with regard to privacy of their data. Such discussions are more engaging for students when there are real individuals to be considered rather than in the context of some imaginary system. A satisfying outcome of these discussions was that a number of security measures were implemented in both the database as well as the application that interfaced with it. Two aspects of security were analyzed: system security and data security. System security refers to the access and use of the database at a system level, such as using a username and password while data security refers to access and use of database objects and the associated actions available on each object [15]. Design issues such as connection traps [2, 4], which are design flaws that can lead to misinterpretation among data relationships, were identified and corrected. We opted to use the PostgreSQL database management system (DBMS) since it is open source with a large, active community of developers. This assured us of long-term stability of the DBMS. More important is the fact that PostgreSQL has a number of sophisticated enterprise features [13], missing from many other open source DBMSs, but essential for a secure and robust environment. A number of queries, including several fairly complex ones, were designed to enable retrieval and presentation of data in meaningful ways. This process was particularly powerful pedagogically since the students were obligated not only to gain a good understanding of the journalistic needs but also to seriously consider how the data could best be presented so that it would be compelling for the journalists and their audience. It was clearly this aspect of the system that was most valuable to the journalists and eventually contributed to convincing the editors of the Philadelphia Inquirer. Access to the database was provided through a web-based application that utilized a number of web technologies and programming languages such as PHP, HTML, CSS, and JavaScript. PHP [12] was especially suitable for this project because it has a number of built in libraries tailored to web programming, including those for addressing web security issues [16] such as SQL injection, cross-site scripting, and session hijacking [17]. Additionally, there is native support for the objectoriented paradigm and its ability to interface with PostgreSQL. The application exposed interface functionality such as helpful display messages, drop down menus, radio button selects, form validation, and clean layout, all designed to help the journalists conduct their investigation effectively. A sample screenshot of Gumshoe is shown in Figure 2. Being aware of the severe newsroom cutbacks in our region, we decided the time was right to approach potential newspaper partners and propose the use of our students’’ analysis in a major journalistic endeavor. While we considered approaching at least three different newspapers in our region, we settled on The Philadelphia Inquirer because of our connections there (two of the authors of this paper are former Inquirer reporters) and because, unlike in our state of New Jersey, Pennsylvania’’s court dockets are online, making them more easily accessible.

4.2 Database design The conceptual design of a database focused on constructing a model of the data, i.e. the entities and their relationships, independent of all physical considerations, such as operating systems, database management systems, computer platforms, and other hardware [2, 4]. The resulting entity-relationship diagram (ER diagram) is shown in Figure 1.

Figure 1: ER Diagram for Gumshoe

532

Taking a broad look at the crimes of murder, rape, robbery and aggravated assault, the newspaper revealed a dysfunctional system in which defendants had a better than 50-50 chance of walking free, with dismissal or withdrawal of all charges. The series showed that Philadelphia had the nation’’s lowest felony conviction rate even as law-enforcement authorities struggled with crime rates above the national average. They also identified other problems, extensive witness intimidation, thousands of fugitives with unresolved cases, and approximately $1 billion in uncollected bail [11]. The reporters (including one of the authors of this paper, who at the time was working at the Inquirer while teaching at our institution) sought to show the depth of the dysfunction in Philadelphia’’s overloaded criminal courts by focusing on violent crimes committed with guns. In nearly twothirds of violent-crime cases, the reporters found, the defendants walked free, and only 20 percent ended with felony convictions [11]. Most of the cases died in Municipal Court, despite the fact that prosecutors must present only probable cause that a defendant committed a crime to allow a case to proceed to trial in the Court of Common Pleas. The newspaper examined 31,000 criminal cases that involved more than 250,000 individual charges –– a targeted analysis of the more than five million records obtained in a huge data run from the state Administrative Office of Pennsylvania Courts, which compiles statewide court records. With a history and experience of publishing major investigative series about problems in the Philadelphia courts, the Inquirer decided, as had we, to examine the current state of the system by looking at how serious cases involving guns, which have become astonishingly available in many neighborhoods of the city, were handled in the city court system. The Inquirer found that prosecutors were most successful in murder cases involving guns, with 83 percent of defendants convicted of at least one charge. But the guilty verdicts plummeted for three other crimes in which guns were used –– rape, robbery and aggravated assault. The most stunning conclusions involved robberies and assaults committed with guns. In robbery cases, just 35 percent of defendants were found guilty of any charge and only 19 percent were convicted of robbery and a gun charge. In aggravated assault cases, 35 percent were found guilty of any charge and 10 percent were found guilty of aggravated assault and a gun charge [11].

Figure 2: Screenshot of Gumshoe

5. ANALYSIS At the end of the semester, we presented the data and results of our analysis to The Inquirer. Based on 859 dockets in a twomonth sample, some of the conclusions we reached were: 1. 392 (45%) cases were withdrawn by the district attorney or discharged by the judge at the preliminary hearing level. At least 55 cases (6%) were nolle prossed in Common Pleas Court. Overall, this meant that at least half of all these prosecutions were dropped or discharged. 2. In nearly four cases out of 10, civilian witnesses and/or victims failed to appear one or more times at court hearings. 3. Only 18 (2%) resulted in convictions on all charges. 4. 162 (19%) resulted in prison time. So, in our sample, the chance of going to jail after being arrested for possession of an illegal firearm and/or violent felonies was less than one in five. The vast majority of defendants received jail time only if they committed violent felonies in addition to gun possession. 5. Among the 162 cases, few received the so-called mandatory minimum of five years in prison for committing a violent felony with a gun. In part, that was because the more serious F1 felonies frequently were reduced to lesser degrees, dropped or plea bargained away. In fact, among those 162 cases, only 60 (37%) resulted in minimum sentences of at least five years. That represented less than 7 percent of the total dockets. 6. In examining the 60 dockets in which there were sentences of five years or more, we found that they involved 38 defendants. (For many incidents there were multiple dockets.) Thus 6 percent of the defendants received jail sentences of five years or more.

6. OUTCOMES AND CONCLUSIONS The student and Inquirer projects, while carried out independently of one another, followed similar paths and, more important, reached virtually identical conclusions. This demonstrates that collaboration between carefully supervised students and veteran journalists in the field can be effective. In fact, the students reached their conclusions well in advance of The Inquirer because two of the staff writers had to put aside the project to cover a five-month trial in late 2008 and early 2009. The Gumshoe project provided a compelling avenue to teach database design concepts as well as issues relating to privacy and security. Having a real ““client”” the students could interact with to obtain requirements and get feedback from, as well as the notion of civic engagement, motivated them to keep on task and to be more thoughtful in the design and development not only of the database but also the user interface and application. In doing so, they gained a deeper understanding of and internalized database concepts such as data modeling, queries, normalization, transaction and concurrency control, as well as issues of privacy and security of data. By working with students or ““clients”” from another discipline, students better understood the techniques and

While our analysis convinced the editor of the Philadelphia Inquirer that this issue needed to be exposed, he was skeptical about the abilities of an undergraduate cohort. The newspaper conducted its own, more sophisticated and broader computer analysis of some of the same data compiled by our students, but ultimately reached nearly identical conclusions. In December 2009, the Philadelphia Inquirer published a four-part series that documented the widespread failure of violent-crime prosecutions in the Philadelphia court system [11].

533

process for effectively eliciting requirements for a system outside their familiar domain as well as for working together as a team to complete a project on schedule. The journalism and computer science students also participated in professional discourse on issues of privacy and security related to the storage and use of private information of the alleged criminals. The investigative series in the Philadelphia Inquirer [11] is resulting in an overhaul of the justice system in that region. Comments we received from some of the students who participated in this project (now all graduated) include: x ““The gun project was the best lesson I got in the value of public records and how they can be used to unlock stories. It made me realize that any time you have access to a collection of public records, you're sitting on a potential gold mine.”” x ““I found this gun project a superb thing to be part of. For one, it improved my news gathering and research skills. Now, I do not just read other news reports for story ideas but make sure to look at data and figures. It is something I am using now as I write my MA dissertation. I am always asking myself: what do the figures say about this particular human rights issue?”” x ““I’’m ecstatic that Gumshoe was used to uncover key defects in the prosecution system and publish these findings to the world. It's always wonderful to hear how your work is benefiting others!””

the Scope and Nature of Computational Thinking. The National Academies Press. [2] Connolly, T., and Begg, C. 2005. Database Systems: A Practical Approach to Design, Implementation, and Management. 4th Ed. Harlow, England: Addison-Wesley. [3] Downie, L., and Schudson, M. 2009. The Reconstruction of American Journalism. The Journalism School, Columbia University, New York NY. [4] Elmasri, R. and Navathe, S.B. 2007. Fundamentals of Database Systems, 5th edition. Addison-Wesley. [5] Holovaty, A. 2006. "A fundamental way newspaper sites need to change." http://holovaty.com/blog/archive/2006/09/ 06/0307/?highlight=structured%2520data. Accessed March 31, 2010. [6] Mecklin, J. 2009. Deep Throat Meets Data Mining. MillerMcClune Jan/Feb 2009, 1, pp 8-13. [7] McGinty, J. C. 2005. "Breaking Down Hate Crime." The New York Times July 24 2005. http://query.nytimes.com/gst/ fullpage.html?res=9C03EFDC153FF937A15754C0A9639C8 B63&scp=1&sq=Breaking+Down+Hate+Crime&st=nyt. Accessed March 31, 2010. [8] Murray, J. H. 1997. Hamlet on the Holodeck, The Future of Narrative in Cyberspace, MIT Press, Cambridge MA.

Gumshoe is a model not only for collaboration with a professional news organization, but also for how students at a small, primarily undergraduate liberal arts institution can make a substantial contribution toward an important but endangered kind of journalism: investigative reporting. Students came away from this project with enhanced analytical skills and a deeper appreciation for the type of computational thinking that is so critical as the United States seeks to preserve its competitive edge in the global market. At a time when the economic crisis in the news industry threatens the ability of news organizations to conduct time-consuming, detail-oriented analyses, it is especially important to continue exploring such models of collaboration.

[9] Niles, R. 2006. The Journalist as Programmer: Q & A with Adrian Holovaty. Online Journalism Review June 5, 2006. http://www.ojr.org/ojr/stories/060605niles/index.cfm. Accessed March 31, 2010.

7. FUTURE STEPS

[12] PHP Manual. 2010. http://www.php.net. Accessed March 31, 2010.

[10] Pew Project for Excellence in Journalism. 2010. The State of the News Media: An Annual Report on American Journalism. http://www.stateofthemedia.org/2010/index.php. Accessed March 30, 2010. [11] Philadelphia Inquirer, The. 2009. Justice: Delayed, Dismissed, Denied. http://www.philly.com/philly/news/ special_packages/79211302.html. Accessed March 30, 2010.

Our journalism and computer science programs intend the Gumshoe collaboration to be the first of many such projects involving local newspapers, most of which have suffered newsroom budget and personnel cuts. When practicable, we intend to incorporate these projects into our regular journalism classes. We did not do this with Gumshoe, choosing instead the independent-study route, because it would have required far too much class time. Today, as the notion of outside collaborators is becoming more acceptable in newsrooms throughout the country, we are formulating plans to approach additional newspapers with the intent of establishing partnerships.

[13] PostgreSQL. 2010. ““PostgreSQL: The world's most advanced open source database. PostgreSQL Global Development Group. http://www.postgresql.org/. Accessed March 31, 2010. [14] Pulimood, S. M. and Wolz, U. 2008. Problem Solving in Community: A Necessary Shift in CS Pedagogy. In Proceedings of the 39th Technical Symposium on Computer Science Education (ACM SIGCSE), Portland, Oregon, March 12 –– 15, 2008. [15] Schach, S. R. (2007). Object-Oriented Classical Software Engineering. 7th ed. New York, NJ: McGraw-Hill.

8. ACKNOWLEDGMENTS This project would not have been possible without the dedication and efforts of our students in the Database Systems and Computer Assisted Reporting independent study classes.

[16] Shiflett, C. (2005). Essential PHP Security. Sebastopol, CA: O’’Reilly Media.

9. REFERENCES

[17] Welling, L. and Thompson, L. (2005). PHP and MySQL Web Development. 3rd ed. Indianapolis: Sams Publishing.

[1] Committee for the Workshops on Computational Thinking, National Research Council. 2010. Report of a Workshop on

[18] Wing, J. M. 2006. Computational thinking. Commun. ACM 49, 3 (Mar. 2006), 33-35.

534