Thomas Lancaster

A Comparison of Source Code Plagiarism Detection Engines.

Automated techniques for finding plagiarism in student source code submissions have been in use for over 20 years and there are many available engines and services. This paper reviews the literature on the major modern detection engines, providing a comparison of them based upon the metrics and techniques they deploy. Generally the most common and effective techniques are seen to involve tokenising student submissions then searching pairs of submissions for long common substrings, an example of what is defined to be a paired structural metric. Computing academics are recommended to use one of the two Web-based detection engines, MOSS and JPlag. It is shown that whilst detection is well established there are still places where further research would be useful, particularly where visual support of the investigation process is possible.

Plagiarism issues for higher education

Academic institutions are finding they have to operate under a pro‐active anti‐plagiarism policy, where plagiarism is actively sought out as a serious breach of acceptable academic behaviour. This paper considers the reasons that institutions need such a policy and the issues they should be aware of when implementing one.

Classifications of plagiarism detection engines

Abstract Plagiarism detection engines are programs that compare documents with possible sources in order to identify similarity and so discover student submissions that might be plagiarised. They are currently classified as either attribute counting systems or structure metric systems. However these classifications are inconsistently applied and inadequate, an area of particular relevance when classifying free text detection engines as this is where most current research is focused. This paper proposes a new and alternative set of classifications based primarily around the types of the metrics the engines use. Current detection engines are classified using the new groupings. These new classifications are intended to allow detection engines to be discussed and compared without ambiguity.

Establishing a Systematic Six-Stage Process for Detecting Contract Cheating

This paper proposes a systematic six-stage process that tutors can use to detect students who are contract cheating. Contract cheating is where students have original work completed for them and submit it, without acknowledgement, for academic credit. A background to the problem is presented along with the current problems of preventing and detecting contract cheating. Examples of how pervasive computing techniques have made it easier for students to cheat are given. A description of how contract cheating is currently detected is presented, with shortcomings of these methods detailed. The paper formalises new six-stage contract cheating detection process developed to parallel approaches taken in the plagiarism literature.

Preserving academic integrity—fighting against nonoriginality agencies

Blackwell Publishing, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA. British Journal of Educational Technology Vol 38 No 1 2007 153–157 doi:10.1111/j.1467-8535.2006.00491.x Blackwell Publishing Ltd.Oxford, UKBJETBritish Journal of Educational Technology0007-1013British Educational Communications and Technology Agency, 20052005381153157Colloquium ColloquiumBritish Journal of

Towards an error free plagarism detection process

For decades many computing departments have deployed systems for the detection of plagiarised student source code submissions. Automated systems to detect free-text student plagiarism are just becoming available and the experience of computing educators is valuable for their successful deployment.This paper describes a Four-Stage Plagiarism Detection Process that attempts to ensure no suspicious similarity is missed and that no student is unfairly accused of plagiarism. Required characteristics of an effective similarity detection engine are proposed and an investigation of a simple engine is described. An innovative prototype tool designed to decrease the workload of tutors investigating undue similarity is also presented.

Commercial aspects of contract cheating

The process of contract cheating, the form of academic dishonesty where students outsource the creation of work on their behalf, has been recognised as a serious threat to the quality of academic awards. Unlike student plagiarism, this cheating behaviour is not currently detectable using automated tools. This paper analyses the monetary value of contract cheating to the different parties who play a role in the contract cheating process. The main analysis is based on a corpus consisting of 14,438 identified attempts to cheat. The corpus was collected between March 2005 and July 2012. The corpus was formed as part of a manual contract cheating detection process identifying students using online agencies. These online agencies are web sites which enable students to contract cheat. The agencies usually benefit from this by receiving a percentage cut of the money raised from the contract cheating that they facilitate. This corpus is used as the basis of an attempt to quantify the monetary value of contract cheating to online agencies. Other parties exist who benefit from the contract cheating process. The paper identifies several such parties and gives examples of the monetary value of contract cheating to each of them. Most notably this includes the contractors who bid for the opportunity to produce work on behalf of the students. Further, the paper identifies the role of intermediary contractors. These are people who post assignment requests on agency sites but who are not themselves students. These intermediary contractors appear to benefit by first receiving requests to complete work for students and then re-outsourcing this work at a much lower cost than they were paid. The group of frequent workers, that is people who regularly work on student assignments and hence benefit financially, is also identified. The paper concludes by presenting the changing trends in contract cheating that the authors have observed since they started working against this form of academic misconduct in 2005. Finally, recommendations for academics towards dealing with the issues posed by contract cheating are provided.

Contract Cheating: The Outsourcing of Assessed Student Work

This chapter reviews the issues associated with contract cheating, loosely defined as the outsourcing of student work to third parties. The chapter is intended to serve as an overview of current research, whilst also providing practical instruction and guidance to academics and educators. The discussion begins by introducing contract cheating and comparing this specific form of academic misconduct with student plagiarism. The suggestion is made that current anti-plagiarism methods are not suitable for contract cheating, defined as where a student is requesting an original bespoke piece of work to be created for them. Six types of services that students can use to have work produced for them are listed; these are: (1) essay writing services, (2) friends, family and other students, (3) private tutors, (4) copyediting services, (5) agency websites, (6) reverse classifieds. Specific challenges associated with each service are provided. Findings related to the wider contract cheating area are given. This includes particular discussion of the research into agency sites, where a student makes the offer of work available to large number of people who then bid to complete it. The student selects one of the bidders to complete the work based on a form of cost-benefit analysis. Issues considered include the extent of contract cheating, the cost and quality of outsourced work and the range of subjects in which students are cheating. The chapter concludes with a discussion of the main methods that can be used to prevent contract cheating. Research into technical solutions, such as stylometrics, put in place to find automated technical solutions to detect contract cheating, is also presented.

An Initial Analysis of the Contextual Information Available within Auction Posts on Contract Cheating Agency Websites

The advantages of using contextual information in order to detect contract cheating attempts by students have not yet been fully explored in the academic literature. Contract cheating occurs when a student uses a third party to produce assessed work for them. This paper focuses on contract cheating using agency websites, where an auction type process is used by students to select a contractor to have the assessed work produced for them, often at a financially advantageous price. Currently, the process of finding contract cheating on agency sites is manually intensive, with a detective required to investigate and attribute each cheating attempt. This paper aims to formally identify the context internally and externally available for contract cheating posts on an agency website. The paper is offered as a starting point for academics interested in producing an automated intelligent contextually-aware tool to detect contract cheating.

The Application of Intelligent Context-Aware Systems to the Detection of Online Student Cheating

Student cheating and plagiarism present a combined threat to the value of academic awards. The technological age has increased the ease with which students can cheat. Although some computerised solutions exist to detect plagiarism and cheating in its other forms, many of these are easily fooled. For other types of cheating, technical solutions are not yet widely available. This suggests that students are receiving awards that they do not deserve. This paper presents three different examples of student cheating, all of which provide academics with a problem. These issues include: (1) plagiarism of documents through automated essay spinning, (2) social media facilitated student cheating groups, and (3) contract cheating using agency and auction web sites. Each of these problems is shown to benefit from the provision of intelligent context-aware systems. The context behind each problem is explored and suggestions for technical implementations of intelligent context-aware systems for each are provided.