Sunteți pe pagina 1din 29

Research Ethics: A Computer Scientists Perspective

By: Usman AHMED LIRIS, INSA de Lyon

Introduction

Science has made a huge contribution in changing our lives from early stone-age to modern technological era. All this was made possible because of extensive research in science and technology. For all this scientific development, the scientists had to made efforts to collect and analyze data, record and document the research, conduct experiments, produce data, propagate the obtained results and observations etc. To facilitate the propagation, researcher and publishers came up with the idea of scientific journals, conferences and seminars. The conferences and seminars facilitate the direct interaction among the research community while the journals are aimed at propagating the research in form of scientific articles. These journals normally have a review committee constituted of the acknowledged researcher in the related fields who are responsible for accepting or rejecting a submitted article for the publication. All these activities that are involved in scientific research raises some ethical issues. To address such issues different organizations and scientific institutions have issued guidelines to conduct a responsible research and have developed some standards. According to the joint NHMRC / AVCC Statement and Guidelines on Research Practice (1997) [2]: The processes of research protect the truth. Communication between collaborators; maintenance and reference to research records; presentation and discussion of work at meetings of experts; publication of results, including the important element of peer review; and the possibility that investigations will be repeated or extended by other researchers, all contribute to the intrinsically self-correcting and ethical nature of research. The guidelines issued by the NHMRC/AVCC address the issues that are common to all scientific disciplines and include the guidelines on different processes of scientific research and provide an overall view about the ethical conduct of the scientific research. But the employed research methodologies and overall research process can be quite different across the scientific fields. For example, in biological and chemical sciences researchers involve in laboratory experiments and work with different apparatus, properties of different mixtures, their testing on subjects (animal and/or human) and reporting the obtained results based on their initial hypothesis which helps in understanding the nature/properties/ problems related to tested compounds, their usefulness and/or side effects etc. The collection of data for the research in these fields may also include the sample collection from animal and/or human population.

In disciplines related to humanities and social sciences, research is generally based on the proposition of some model or theory, its implementation in a population, research surveys of a population, characterizing the survey results on the basis of different population characteristics, retrieving interesting results using statistical techniques determining the populations behavior (on the basis of sample), change in the behavior over the time and/or caused by some specific events etc. This type of research includes the interaction of researcher with human subjects, collecting information about their private life or their likes and dislikes etc. On the other hand, in computer science, the research is related to providing novel solutions for data storage and acquisition, privacy of stored data, efficient data retrieval, information retrieval techniques, machine learning, solutions to easily understand and visualize the data, simulations of different phenomena, artificial intelligence, data matching techniques etc. The data for computer science researchers could come from other fields or could be gathered through social networking and other websites, online games etc. The use of self generated synthetic data is also very common in computer science research. As we have seen, different disciplines deal with different types of problems. Therefore, the scientists working in these fields also have different perspectives on the issues related to classical scientific research. In this report we discuss some of the ethical issues related to scientific research and discuss how it could be different for a computer scientist. We discuss the reasons that could lead a computer scientist to have different perspective on classical ethical issues related to following research processes. Publications and Authorship Research Supervision Research Collaboration

We also discuss the following issues that could arise during the research process. Conflict of Interest Intellectual Property Intellectual Misconduct

At the end of the report we include self-explanatory results of the survey conducted on these issues on the population involving only the computer scientists.

Publications and Authorship

The publication of obtained research experimental results and collected data by the research community is important to further the research. Many publishers started publishing the research work after reviews by expert in the concerned fields. With the time these research publications have become a metric to evaluate the research capabilities of a researcher. Many academic, research and other institutes and organizations started to use this criterion to rank the researchers during their recruitment and evaluation process. Ranking mechanisms (based on impact factor etc.) of research journals are developed and the researchers also try to publish in the journals with better impact factor. Other metrics like h-index, citation-count etc. are also used to evaluate the researchers capabilities. This ranking and evaluation criteria increases the competition among the researcher which on one hand could be useful for scientific development but on the other hand it could be drastic as it might encourage some researchers to include the names of the colleagues just to increase the chances of paper acceptance or increasing the colleagues number of publications. Then order in which the names of author should appear on a publication is also important in some fields. Like in biotechnology the general rule is to use put then names as first author the main contributor, last one the research group super visor etc. In mathematical sciences, the alphabetical order is used while in computer science and related discipline the order of the authors reflects the contribution of the researchers and research group supervisor's name is generally not included in the paper if he does not contribute. Participation in acquisition of funding might also encourage the researchers to include the names of people who didn't actually contribute. Researchers some time include the name of authors who just reviewed or gave some suggestion. Some researchers suggest adding such names under a special acknowledgment section while other might simply ignore their names. While publishing some results or data, should a researcher take the permission of all the people whose names appear as author? This permission should be written or oral permission is ok? Do all the authors need to review the final article and are they ready to take the responsibility of whatever is being published? The decision about the timing of a publication is also important. The researchers must decide among them that when to publish the data and/or results? Which journal should be selected for the publication and which parts of the research should be published?

Another issue that may rise while publishing a research article is how to include the name of affiliations? If a researcher can add the name of an organization to which he is not practically affiliated and/or which did not participated in the research activity? All the above discussed issues are common to all the scientific disciplines, but in computer science, normally a publication has very few authors as opposed to the publications in other natural and physical science where some works may have a large number of authors. The reason for having a large number of authors in those disciplines is that large research projects are split into many small interconnected parts. Different research laboratories or groups work on different parts and all of them contribute to get the publishable results. On the other hand, large computer science projects are generally split into small independent parts. The results obtained for one part are independent (and publishable) of the other. So a huge project can have many contributors but who are independent of each other and hence publish the results independently. So a computer scientists perspective on some of these issues might be different than the researchers working in other scientific disciplines.

Research Supervision

To guide young researchers, the mechanism of research supervision is adopted. In the research fields where the research includes working in the laboratory, the supervisor is responsible for explaining the research conduct, recording of experimental details and the results, maintenance and cleaning of apparatus after the use, keeping the working environment clean, security measures and maintaining the laboratory books. He might also need to explain the working of apparatus, reading and manipulating the results etc. Some apparatus are being used by many people in the lab which may need the reservation, scheduling the experiments etc. which also needs to be explained to the young researchers. In computer science, most often the work is done on a computer with possibly linked servers. But most of the experimental process is done on the personal computers. The researchers are given instructions about the working environment and fair use of computer resources. The researchers are given instructions about writing a computer program. This includes the possible choices of programming languages, adherence to standards of the laboratory, documentation and comments about written programs and the backup strategy.

The supervision may also include the guidance on selection and reading of the papers from literature. These instructions may describe how to read a paper and how to maintain the bibliography and the important points in the read paper for later reuse (summary of the article etc.) So the research supervision is almost similar to all the disciplines with no special difference for computer scientists.

Research Collaboration

Research collaboration has now become very common among many researchers. This collaboration could be among various research groups, different research institutes of the same or different countries, the researchers from different fields or cultures. The aim of such collaboration is to benefit from the expertise and resources of different collaborators and to accelerate the research process. This, on hand is very useful for the advancement of the research while on the other hand it could raise some issues among the collaborators. In natural and physical sciences, it is quite common to have such collaborations. The research groups very often work on large projects that are split among many small research groups that work on their own parts but may depend on the results from other groups. In computer science, however, the projects are very often small. These projects, therefore do not involve as many collaborators as in other fields. But, since the computer science is involved almost in every field of life, the interaction of computer science researchers with the people from other research communities or other walks of life is much more common. This interaction may involve the interaction with both technical and non-technical people who have completely different perspectives of different issues. So, due to the different natures of collaborations (or collaborators) among different fields, there is a possibility that people coming from different fields have a different perspective and approach towards the issues and challenges coming during the entire collaboration process.

Conflict of Interest
Conflict of interest could be defined as a situation in which financial or other personal considerations have the potential to compromise or bias professional judgment and objectivity [3]. It is important to understand that conflict of interest does not necessarily mean that the researcher has violated the ethical research of conduct (e.g. committing plagiarism etc.). It only implies that it could possibly

motivate or encourage a researcher to compromise the actual objectivity of the research work to gain financial, professional or other personal benefits. Due to existing conflict of interest, a reviewer may respond positively to a low quality manuscript or could reject a better quality manuscript just because he has some personal or professional inclination or grudge for the author(s) of the manuscript. Another possible situation may exist where a researcher comparing different products could misreport the obtained results to highlight the product of his funding organization or could understate the results of another competitor. Therefore, it is necessary for an ethical conduct of research that any possible conflict(s) of interest be declared at the time of publication of data and/or results. As the conflicts of interest seem to be same for all the scientific fields, we do not expect much of difference among the point of view of computer scientists and those coming from other scientific fields.

Intellectual Misconduct

As discussed, the publications are used as evaluation criteria which may be used by the academic and research institutes for the recruitment and/or promotion process. This puts the pressure on researchers to produce more and more publications. But as the publication process is very competitive, this may encourage the researcher to become the culprit of intellectual misconduct. The intellectual misconduct for research includes, but is not limited to: Plagiarism Falsification of data and/or analytical process Lying about the experimental setup and/or methodology Fabrication of data Omission of deviating results Omission of already published work by himself or other researcher Breach of confidentiality by a reviewing researcher

In other scientific fields, one needs to provide all the details and results reproduction is easier. However, in computer science, metrics are not well defined and the data falsification is much easier [1] because the comparisons could be made using different metrics and software. Indigenously developed softwares are not generally shared with other researchers. The sharing of source code of the

developed program or software is much rarer. This prevents the research community to verify if some intentional misappropriations are done in the software to get the desired results. The level of memory management and use of different data structures could also lead the comparisons to be unrealistic. To deal with such problems, some benchmarks are introduced but the use of entire benchmark in performance evaluation is very rare. For example, The Standard Performance Evaluation Corporation (SPEC) CPU2000 [4] suit of benchmarks is an industry-standardized reference for measuring a computer systems processor, memory and compiler support [ref paper page 2]. In [Citron03], the author found that out of 115 research papers published in highly acknowledged journals and conference that claim using the SPEC2000, only 23 used the entire suite. Such partial use of benchmarks in computer science is very common, which could not be thought as sufficient for drawing realistic comparisons among different solutions and hence could be misleading. These differences in different scientific fields may lead the computer scientists to think differently than the scientists of other fields.

Intellectual Property

The term intellectual property refers to the exclusive rights held by a researcher on creation of his mind or proposed ideas. As the research work has many contributors (research lab, financing agency etc.), it is important to decide that who holds these intellectual property rights and to what extent. Should the detail theoretical and experimental data and results be recorded in a readable format to ensure that the researchers working afterwards in your laboratory could understand what you did, even if the experiments did not produce the desired results? In computer science related research, laboratory books are not generally maintained. Data is kept on the disk in electronic format and in general practice does not include the detailed description of each and every step carried out on daily basis. The source code of programs or software written for performance evaluation however may need proper documentation, comments and adherence to the defined standards (if any). So that the source code could be easily reusable after the researcher has left the research group. The perspective on this issue is also expected to be different for computer scientists than those who come from other scientific fields.

Conclusion

From the above discussion, we can conclude that ethical conduct of research requires certain aspects to be taken into the consideration. Most of these aspects are common fro all disciplines, but due to some differences in the working environment and the methodology employed in different disciplines, the perspective of the researchers working in different fields could be different. In this report, we discussed some of the possible differences that exist between the research in the field of computer science and other natural and physical sciences. These possible differences may affect the perspective of these computer scientists. Hence, we carry out a research survey to know there perspective on different issues related to research ethics and present the summarized results in the following. The presented results are self-explanatory; therefore we do not provide the comments.

References

1. David R. Wright, Research Ethics and Computer Science: An Unconsummated Marriage, In Proceedings of SIGDOC06, Myrtle Beach, South Carolina USA, October 18-20 2006. 2. Justin Zobel, Guidelines on Research Practice in Computer Science,

http://goanna.cs.rmit.edu.au/~jz/conduct.html (accessed on 20th May, 2012), Australia, May 1999. 3. RCR Conflicts of Interest : Responsible Conduct of Research,

http://ccnmtl.columbia.edu/projects/rcr/rcr_conflicts/foundation/index.html, accessed on 20th May, 2006. 4. Standard Performance Evaluation Corporation. CPU2000. http://www.spec.org/cpu2000/ (accessed on 20th May, 2006), 2000.

Participants
The questionnaires were translated from French to English which were originally provided in the class rooms of a course titled Ethique, Intgrit Scintifique et Bonne Pratique du Laboaratorire. The survey was conduct online using GoogleDoc spreadsheet. The invitations (and reminders) were sent only to computer scientists with clear mentioning of Only for Scientists working in the Field of Computer Science. Following is the distribution of scientists according to their status who filled the survey and provided their feedback.

PhD Students Post-doc Assistant Professor (MCF sans HDR) Associate Professor (MCF avec HDR) Professor Other - Researcher

19 1 3 0 0 2 Total: 25

Research Publications

Two months after joining a new research group, you are ready to submit an article covering the work done before your integration into the group. Dr. Helix one of your new colleagues recommends that you include among the authors, the name of Dr. Spiral, in charge of your new research group. You make it out that Dr. Spiral did not contribute to the work described, but Dr. Helix replies that the addition of Dr. Spiral's name will increases the likelihood of the paper acceptance and improves your chances of promotion within his group. Based on this information, would you agree to include the name of Dr. Spiral in your paper?

Yes No Dont Know

5 9 10

20% 36% 40%

Is the addition of Dr. Spiral's name ethical?

Yes No Dont Know Being part of the authors of a scientific article is justified if the person: actively participated in the experimental part gave the idea of a crucial experiment for obtaining the described results gave crucial biological material for the experiments described in the paper carried out an experiment using a small but highly specialized equipment gave unpublished data in addition to results presented in the paper

2 18 13

8% 72% 12%

22 88% 18 72% 13 52% 18 72% 18 72% 16 20 5 16 0 7 5 64% 80% 20% 64% 0% 28% 20%

did experiments cited in the paper but do not appear as figures conducted the statistical analysis associated with the results presented in the paper contributed to the paper only in the form of services wrote some of the paper but did not conduct experiments included in the paper seeks only to advance his career is responsible for the team which was carried out scientific work has only read and commented on the latest version of the manuscript

You do a postdoc in Dr. Strauss's Lab. Dr. Strauss was asked to review a manuscript, she asks your opinion on this manuscript. You ask her if she has told the publisher that you would also review the manuscript. She responds that it is not necessary because it is common to share a manuscript for review.

Based on this information, would you give your consent to review the manuscript?

Yes No Dont Know

15 5 4

60% 20% 16%

Do you think not informing the editor is unethical?

Yes No Dont Know Without informing the editor, the "reviewer" can:

8 10 5

32% 40% 20%

Share the content of a manuscript with a student or colleague in order to get help

Yes No Dont Know

8 10 5

32% 40% 20%

Share content of the manuscript with others to teach them how to write an expertise

Yes No Dont Know

7 15 1

28% 60% 4%

Share content of the manuscript with others to inform them of the latest

Yes No Dont Know

4 18 1

16% 72% 4%

Use the results presented in the manuscript for his own work before the publication of the manuscript.

Yes No Dont Know

0 22 1

0% 88% 4%

Someone who is asked to appraise a manuscript or a research project in the heart of its own subject must decline the invitation because the risks are too strong bias

Agree Disagree Neutral

6 15 3

24% 60% 12%

Someone who is asked to appraise a manuscript or a research project that has friendly ties with one of the authors of the manuscript should decline because the risk of bias is too strong

Agree Disagree Neutral

16 4 4

64% 16% 16%

Someone who is asked to appraise a manuscript or a research project and has professional links with one of the authors of the manuscript should decline because the risk of bias is too strong

Agree Disagree Neutral

13 5 6

52% 20% 24%

Someone who is asked to appraise a manuscript or a research project which has had serious differences with one of the authors of the manuscript should decline the invitation because the risk of bias is too strong

Agree Disagree Neutral

18 3 3

72% 12% 12%

Research Supervision

Your supervisor is/should be responsible to advise you

Agree Disagree Neutral

22 1 0

88% 4% 0%

Supervisor is/should be directive

Agree Disagree Neutral

15 4 4

60% 16% 16%

The research team leaders must: Impose a probationary period on PhD students before accepting him in his laboratory

Agree Disagree Neutral

7 6 11

28% 24% 44%

Set a maximum number of doctoral students in the team. This number is determined by financial constraints and the working area but also by the leadership capacity of supervisor

Agree Disagree Neutral

19 2 3

76% 8% 12%

Give clear instructions to PhD students on how to keep their workbooks, archiving, data retention and data properties

Agree Disagree Neutral

15 3 6

60% 12% 24%

Establish a clear procedure for scientific publications: preparation of the manuscript, coauthors. This procedure must be provided to PhD students at the beginning of their work

Agree Disagree Neutral

17 2 5

68% 8% 20%

Meet face to face regularly with each PhD student to discuss progress of the research, analyze data, design experiments and set goals and timelines

Agree Disagree Neutral

20 2 2

80% 8% 8%

Schedule meetings of the entire team to discuss the progress of each individual project

Agree Disagree Neutral

20 2 2

80% 8% 8%

Encourage a direct competition between the PhD team

Agree Disagree Neutral

7 13 4

28% 52% 16%

Helping and guiding PhD students in the choice and reading of scientific articles (e.g. reading method)

Agree Disagree Neutral

17 2 5

68% 8% 20%

Be active in the presentation of PhD students in the scientific community through seminars, conferences

Agree Disagree Neutral

20 1 3

80% 4% 12%

Provide advice and support for job seekers in particular at the end of the thesis

Agree Disagree Neutral

20 0 4

80% 0% 16%

Research Collaboration

It is beneficial to enter into collaborations with other teams to do quality research.

Agree Disagree Neutral

23 1 0

92% 4% 0%

Collaboration in research produce little or no problems

Agree Disagree Neutral

9 8 7

36% 32% 28%

Problems arise in a collaboration, if: The work plan is not well defined at the outset of the collaboration

Agree Disagree Neutral

21 1 2

84% 4% 8%

Teams have similar skills (experimental or otherwise)

Agree Disagree Neutral

6 12 6

24% 48% 24%

The teams come from different disciplines

Agree Disagree Neutral The teams come from academic and industrial sectors

6 13 5

24% 52% 20%

Agree Disagree Neutral The teams are international (in countries with different cultures)

4 13 7

16% 52% 28%

Agree Disagree Neutral

6 12 6

24% 48% 24%

To ensure a healthy and fruitful cooperation, researchers must agree on Who is responsible for what

Agree Disagree Neutral A timetable and deadlines for each step

23 0 1

92% 0% 4%

Agree Disagree Neutral

23 0 1

92% 0% 4%

The criteria for signing Publications

Agree Disagree Neutral

20 3 1

80% 12% 4%

The order of authorship on publications

Agree Disagree Neutral

21 2 1

84% 8% 4%

How the research products will be divided if an interesting application is developed

Agree Disagree Neutral

23 0 1

92% 0% 4%

Conflict of Interest

Authors of scientific publications are always asked to indicate their possible personal financial involvement

Agree Disagree Neutral

11 7 6

44% 28% 24%

It is essential to the reviewers of scientific papers to know whether the authors are personally financially involved in research published

Agree Disagree Neutral

7 15 2

28% 60% 4%

It is essential for readers of scientific articles to know whether the authors are personally financially involved in the published work

Agree Disagree Neutral

6 13 5

24% 52% 20%

A financial interest may encourage an author to fabricate or falsify data

Agree Disagree Neutral

11 4 7

44% 16% 28%

A financial interest may encourage you to fabricate or falsify data

Agree Disagree Neutral

5 13 6

20% 52% 24%

All conflicts of interest, not only financial, may encourage authors to falsify (e.g., improve) data

Agree Disagree Neutral All conflicts of interest, not only financial, increase risk of unintentional bias

12 5 7

48% 20% 28%

Agree Disagree Neutral

15 3 6

60% 12% 24%

The need for regular funding requests can lead the authors to lie about their research

Agree Disagree Neutral

12 3 9

48% 12% 36%

Protection against bias due to conflicts of interest can be ensured by: The repetition of experiments by other authors

Agree Disagree Neutral

18 5 1

72% 20% 4%

The logic of scientific arguments

Agree Disagree Neutral

20 0 4

80% 0% 16%

Data obtained by double blind

Agree Disagree Neutral

17 0 7

68% 0% 28%

Intellectual Misconduct

You can publish experimental data that have been created without carrying out the actual experiments

Agree Disagree Neutral

2 20 1

8% 80% 4%

It is acceptable to modify the experimental data to ensure that results are better than they really are

Agree Disagree Neutral

1 21 1

4% 84% 4%

It is acceptable to use different analytical methods for processing data to obtain, by at least one of these methods, statistical significance

Agree Disagree Neutral

14 5 4

56% 20% 16%

You should never take ownership of the words or writings of someone else (without the quote)

Agree Disagree Neutral

20 1 2

80% 4% 8%

You should never take ownership of the data obtained by someone else (without the quote)

Agree Disagree Neutral

22 0 1

88% 0% 4%

You should never take ownership of the ideas expressed by someone else (without the quote)

Agree Disagree Neutral

20 2 1

80% 8% 4%

If you trust your data, it is acceptable to eliminate some conflicting results for the timely publication

Agree Disagree Neutral

10 6 7

40% 48% 28%

If you have confidence in your results, it is acceptable to falsify or fabricate data to accelerate the publication

Agree Disagree Neutral

2 19 2

8% 76% 8%

It is more important to be sure of his results for publication as a request for funding

Agree Disagree Neutral

12 2 8

48% 8% 32%

If you are a witness of facts contrary to scientific integrity by your peers you have to act and communicate them to appropriate persons

Agree Disagree Neutral

14 1 7

56% 4% 28%

If in a paper it is shown that data were falsified, all co-authors are responsible

Agree Disagree Neutral

12 3 8

48% 12% 32%

If in a paper it is shown that data were falsified, all co-authors must have the same penalty

Agree Disagree Neutral

9 6 8

36% 24% 32%

Intellectual Property

The data that you generate through your research project belong(s) You Your Institution Funding Organization Research Group 8 32%

10 40% 5 1 20% 4%

The responsibility of deciding what is published and when (or sharing the results) is owned by: You Your Institution Funding Organization Research Group 14 56% 2 6 2 8% 24% 8%

You are told that sharing the results before publishing them may be needed to advance the research. Do you think this is a good rule?

Yes No Dont Know

5 11 8

20% 44% 32%

The laboratory notebooks should always be written in indelible ink

Agree Disagree Neutral

13 1 10

52% 4% 40%

The research results should be recorded in numbered and unchangeable notebooks

Agree Disagree Neutral

13 1 9

52% 4% 36%

The research results should be recorded in chronological order

Agree Disagree Neutral

14 1 8

56% 4% 32%

The daily notes about the experiments must: Include a date

Agree Disagree Neutral

19 0 4

76% 0% 16%

Include the name of the researcher

Agree Disagree Neutral Include the objective of the experiment

20 0 3

80% 0% 12%

Agree Disagree Neutral

18 0 5

72% 0% 20%

A summary of what has been done

Agree Disagree Neutral

19 0 3

76% 0% 12%

Be countersigned by a third person competent and with a liability to labour

Agree Disagree Neutral

5 4 15

20% 16% 60%

Laboratory notebooks should always be kept in the institution where they were written

Agree Disagree Neutral

10 3 11

40% 12% 44%

When leaving a group, a doctor or a post-doc should take his original notes

Agree Disagree Neutral

8 8 8

32% 32% 32%

When leaving a group, a doctor or a post-doc can take a copy of his original notes

Agree Disagree Neutral

22 0 2

88% 0% 8%

S-ar putea să vă placă și