Demonstrate the ability to conduct ethical research and follow academic integrity standards through proper literature analysis, citation practices and plagiarism avoidance.
Lecture 4 established what plagiarism is and what it costs. This lecture is entirely about the operational skills that prevent it.
Knowing that plagiarism is unethical is necessary but not sufficient. Four competencies do the actual preventing: paraphrasing, citation styles, reference management and detection tools.
Two items are hands-on rather than readable: install a reference manager (Section 5) and practise the five-step method on three technical passages (Section 3.2). The examination tests both as performance items.
1. Learning Objectives
| # | Objective | Bloom's level |
|---|---|---|
| LO1 | Differentiate between acceptable paraphrasing and disguised copying | Understand |
| LO2 | Apply at least two major citation styles (APA, IEEE) correctly | Apply |
| LO3 | Use a reference management tool to organise sources and generate bibliographies | Apply |
| LO4 | Interpret similarity reports from Turnitin/URKUND and act upon them | Analyse |
| LO5 | Formulate a personal anti-plagiarism workflow for thesis and paper writing | Create |
2. The Four Pillars
Research is a cumulative enterprise, and misrepresenting the source of an idea is among the most serious violations of academic integrity. The operational skills that prevent it fall into four interrelated domains:
- Paraphrasing — the art of restating ideas in one's own voice while preserving meaning.
- Citation styles — the formal language that credits sources unambiguously.
- Reference management — the systematic organisation of bibliographic data.
- Plagiarism detection tools — the technological verification layer that confirms originality.
| Term | Meaning |
|---|---|
| Paraphrase | Restatement of text in different words and structure while preserving meaning |
| Citation | In-text reference acknowledging the source of an idea, finding or quotation |
| Reference list | Comprehensive list of all sources cited in a document |
| Similarity index | Percentage of text in a submission matching comparison databases |
| Originality report | Annotated output indicating matched sources and overlap statistics |
Accidental plagiarism is defeated by reference management; mosaic plagiarism by genuine paraphrasing; direct plagiarism by quotation and citation; idea plagiarism by attributing concepts, not only sentences. The techniques are not interchangeable good habits.
Computer science papers frequently include algorithms, pseudocode and architectural diagrams. Copying an algorithm description from a paper and presenting it as one's own contribution constitutes plagiarism even if the words are modified slightly — the intellectual contribution belongs to the original author.
3. Paraphrasing
3.1 What Paraphrasing Is — and Is Not
A common misconception is that paraphrasing means replacing a few words with synonyms while leaving the sentence structure intact. This practice, called patchwriting, is a form of plagiarism because it preserves the original author's syntax and voice without attribution.
A genuine paraphrase:
- Changes the sentence structure completely.
- Uses vocabulary appropriate to the writer's own discipline and audience.
- Maintains fidelity to the original meaning, without distortion.
- Includes a citation, even though the original words have not been used.
3.2 The Five-Step Paraphrasing Method
- Read and comprehend the original passage thoroughly.
- Set aside the original text so it cannot be visually copied.
- Restate the idea in your own words and sentence structures.
- Compare your paraphrase with the original to ensure accuracy and originality.
- Cite the source, even though you have not used the original words.
Step 2 is the load-bearing one. Paraphrasing with the source in view produces patchwriting almost automatically, because the eye supplies the original syntax before the mind supplies an alternative. Note also the ordering of steps 3 and 4: restate first, then verify — checking during composition collapses back into word substitution. Step 4 also guards against the opposite failure, a paraphrase so free that it misrepresents the source.
3.3 Worked Example
“Deep learning is a particular kind of machine learning that achieves great power and flexibility by learning to represent the world as a nested hierarchy of concepts, with each concept defined in relation to simpler concepts.”
“Deep learning is a specific type of machine learning that achieves high power and flexibility by learning to represent the world as a nested hierarchy of concepts, where each concept is defined in relation to simpler ones.”
Critique: only synonyms are changed — specific for particular, high for great, where for with. Sentence structure is preserved, and this would still be flagged by Turnitin.
“By modeling reality through layers of increasingly abstract representations — where each higher-level construct is built upon the lower-level ones — deep learning derives its representational capacity from hierarchical feature composition, distinguishing it from classical machine learning paradigms (Goodfellow et al., 2016).”
Critique: sentence structure inverted, vocabulary domain-appropriate (representational capacity, hierarchical feature composition), meaning preserved, citation included.
Could a reader reconstruct the original sentence's skeleton from mine? If yes, it is patchwriting.
3.4 Quotation, Paraphrase or Summary
| Technique | When to use | Length | Citation |
|---|---|---|---|
| Quotation | When the original wording is uniquely authoritative or stylistically essential | Exact words; usually 40 words or fewer | Required, plus quotation marks |
| Paraphrase | When the idea is important but the wording is not | Approximately the same length as the original | Required (no quotation marks) |
| Summary | When condensing a long passage into a brief statement | Substantially shorter than the original | Required (no quotation marks) |
Rule of thumb for CS theses: reserve direct quotations for definitions, formal theorems and standards language; use paraphrases for algorithm descriptions and prior-work summaries. Paraphrasing a shall clause from an IEEE standard or an RFC can change its normative meaning, so quote it.
Removing quotation marks does not remove the obligation — it changes which obligation applies. The belief that “no quotation marks means no citation needed” is the direct cause of mosaic plagiarism.
4. Citation Styles
A citation style is a standardised format for documenting sources. Choosing the correct style and applying it consistently is a hallmark of academic rigour. In computing the dominant styles are IEEE Editorial Style, APA 7th Edition and, for ACM venues, the ACM Reference Format.
4.1 APA 7th Edition
- In-text, paraphrase: Recent work has demonstrated the efficacy of transformer architectures for low-resource NLP tasks (Devlin et al., 2019).
- In-text, direct quotation with page number: “The model architecture is a multi-layer bidirectional Transformer encoder” (Devlin et al., 2019, p. 4171).
- Reference entry: Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2019). BERT: Pre-training of deep bidirectional transformers for language understanding. Proceedings of NAACL-HLT, 4171–4186. https://doi.org/10.18653/v1/N19-1423
4.2 IEEE Editorial Style
- In-text: Recent work has demonstrated the efficacy of transformer architectures for low-resource NLP tasks [1].
- Reference entry: [1] J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, “BERT: Pre-training of deep bidirectional transformers for language understanding,” in Proc. NAACL-HLT, 2019, pp. 4171–4186, doi: 10.18653/v1/N19-1423.
Note what stays constant across both: author set, title, venue, year, pages and DOI. Only the ordering, punctuation and in-text marker differ. Learning a style is learning a convention, not new content.
4.3 Element Comparison
| Element | APA 7th | IEEE |
|---|---|---|
| In-text marker | (Author, Year) | [Number] |
| Reference ordering | Alphabetical by author | Order of appearance |
| Title case for articles | Sentence case | Lower case except proper nouns |
| DOI format | https://doi.org/… | doi: 10.… |
| Author format | Last, F. M. | F. Last |
The ordering row has a practical consequence: IEEE numbering by order of appearance means that inserting one new citation renumbers everything downstream — which is precisely why manual citation management fails at thesis scale.
4.4 Which Style, and Style Drift
- IEEE — for submissions to IEEE, ACM, Springer LNCS and Elsevier journals, and most B.Tech and M.Tech theses in Indian universities including UPES.
- APA — for HCI, cognitive science, education research and interdisciplinary work.
- ACM Reference Format — when submitting to ACM venues.
A student begins writing in APA for a coursework assignment and switches to IEEE for the project thesis, producing a mixed bibliography. IEEE reviewers reject such a submission outright, and even APA-only evaluators flag the inconsistency.
Resolution: establish the citation style at the start of the document, configure the reference management software to use that style exclusively, and verify the bibliography against the official style guide before submission. Always confirm the style required by your target venue or supervisor before writing.
5. Reference Management
For a B.Tech student managing 30–80 references across a thesis, reference management software is not a luxury — at that scale manual management guarantees errors in ordering, formatting and completeness.
5.1 The Big Three
| Feature | Mendeley | Zotero | EndNote |
|---|---|---|---|
| Developer | Elsevier | Roy Rosenzweig Center (open source) | Clarivate |
| Cost | Free, with optional premium | Free, open source | Paid institutional licence |
| PDF storage | Excellent | Excellent (with web plugin) | Good |
| Word integration | Yes | Yes | Yes |
| Browser capture | Web Importer | Connector | Web capture |
| Collaboration | Private groups | Group libraries | Shared libraries |
| Best for | PDF-heavy workflows | Open-source advocates | Large institutional projects |
For this cohort, either Mendeley or Zotero. Both are free and integrate with Word and LaTeX; the choice matters far less than adopting one before the writing begins. A reference manager installed after the draft exists recovers almost none of its value.
5.2 The Six-Step Workflow
- Capture — install the browser connector (Mendeley Web Importer or Zotero Connector) and save references directly from Google Scholar, IEEE Xplore, ACM DL, SpringerLink and arXiv.
- Organise — create folders by chapter or topic; rename PDFs using a clear convention such as
Author_Year_Keyword_Title.pdf. - Annotate — highlight key passages and add notes directly in the PDF within the manager. This is where prevention actually happens: annotation converts reading into comprehension, and comprehension is the precondition of genuine paraphrase.
- Cite while writing — insert citations from the Word or LaTeX plugin. Never type citations manually.
- Verify — cross-check the auto-generated bibliography against the official style guide; tools occasionally miss publication-type nuances such as the conference-versus-journal distinction.
- Back up — synchronise the library to the cloud weekly. A lost library is a research catastrophe.
5.3 LaTeX and BibTeX
For theses written in LaTeX, reference managers integrate through BibTeX (legacy but universally supported) or biblatex (modern and more flexible). Mendeley and Zotero export .bib files that the compiler processes; the Better BibTeX Zotero plugin generates stable citation keys such as vaswani2017attention.
The double braces around {BERT} stop BibTeX lowercasing the acronym under IEEE style — a small detail that teaches the useful lesson that the tool is a program rather than a formatter.
6. Detection Tools: Turnitin and URKUND
Plagiarism detection software does not detect plagiarism — it detects textual similarity. The interpretation of similarity reports is a human responsibility. A match is not plagiarism; an unattributed match may be.
These tools compare submitted text against three databases:
- The internet archive — billions of continuously updated web pages.
- Published content — journal articles, conference proceedings, e-books.
- The institutional repository — prior student submissions at the same university. This is why copying a senior's thesis is the most reliably detected form of misconduct in a university setting.
6.1 Turnitin
Submissions are broken into overlapping text segments; each segment is matched against the comparison databases; a similarity index is computed as the percentage of matched text to total submission length; and colour-coded highlights link each match to its suspected source. Turnitin is integrated into the UPES learning-management ecosystem.
6.2 URKUND (Ouriginal)
- Cross-language detection — can identify translated plagiarism.
- Modular API — readily integrated into an institutional LMS.
- Character-by-character analysis — finer granularity than n-gram matching.
- Generative-AI detection module — added in 2023.
URKUND was rebranded Ouriginal following a 2020 merger with PlagScan. A typical workflow runs: submission → LMS upload → server → document parsed → compared against the three databases → similarity report generated → delivered to the instructor, often within thirty minutes.
| Feature | Turnitin | URKUND / Ouriginal | iThenticate | Free checkers |
|---|---|---|---|---|
| Market | Academic (student work) | Academic and research | Research and publication | General |
| Colour-coded report | Yes | Yes | Yes | Basic |
| Excludes references | Yes (configurable) | Yes | Manual | No |
| AI content detection | Yes (since 2023) | Yes | Limited | No |
| Cost | Institutional licence | Institutional licence | Per-check | Free tier |
6.3 Interpreting the Index
| Similarity index | Typical interpretation | Recommended action |
|---|---|---|
| 0–10% | Excellent originality | Acceptable for most work |
| 11–24% | Acceptable with proper attribution | Review highlighted sources |
| 25–50% | Substantial similarity detected | Significant revision required |
| Above 50% | Likely unattributed copying | Rewriting and resubmission mandatory |
Most Indian universities, including UPES, permit a maximum similarity index of 10–20 per cent for theses, excluding references, quotations and common methodological phrases. Exceeding the threshold triggers mandatory revision and supervisor review.
A 0 per cent score does not guarantee originality: translated, well-paraphrased and non-textual content — code, diagrams — can evade detection. Conversely, legitimate matches arise naturally from common phrases, boilerplate text and standard methodological description.
6.4 Acting on the Report
- Review each highlighted segment individually.
- Identify the source by clicking on the match.
- Classify each match as: a properly cited quotation (acceptable); a properly paraphrased passage with citation (acceptable); common knowledge or a methodological phrase (excludable); a reference-list entry (excludable); or unattributed textual overlap (action required).
- Revise the offending passages using the five-step method.
- Resubmit and verify that the index has dropped below the institutional threshold.
Investigation reveals: 8% from properly cited quotations; 12% from methodological phrases common to all CS theses (“the experimental setup was designed to evaluate…”); 5% from reference-list titles (excludable); and 2% from one uncited paragraph lifted from a survey paper.
Effective similarity index: 2 per cent. The thesis is otherwise in good standing, but the student must rewrite the uncited paragraph with proper attribution. Always compute the effective index, excluding acceptable categories — tools provide filters for exactly this purpose, but they must be configured.
7. Case Studies
7.1 The Patchwriting Trap
A student writing the Related Work section of a thesis on federated learning reads McMahan et al. (2017) and drafts by replacing some words with synonyms.
Original: “Federated learning is a machine learning setting where many clients collaboratively train a model under the coordination of a central server, while keeping their training data decentralized.”
Student draft: “Federated learning is a setting of machine learning where many clients train a model collaboratively under the coordination of a central server, while keeping training data decentralized.”
Diagnosis: sentence structure preserved, only minor synonym substitution and reordering. This is patchwriting and will be flagged.
The federated learning paradigm distributes model training across numerous client devices, each retaining local data, while a central coordinator aggregates updates without direct access to the underlying datasets (McMahan et al., 2017).
7.2 Citation Style Drift
A student begins in APA for coursework and switches to IEEE for the project thesis; the bibliography contains mixed entries. IEEE reviewers reject the submission immediately, and even APA-only evaluators flag the inconsistency. The resolution is structural: fix the style at the outset, configure the reference manager accordingly, and verify against the official manual before submission.
8. Your Personal Anti-Plagiarism Workflow
This is the answer to LO5, and to the ten-mark constructive question:
- Read and annotate in the reference manager — defeats shallow comprehension.
- Close the source — defeats patchwriting.
- Draft the paraphrase in your own structure.
- Insert the citation from the plugin, in the mandated style — defeats numbering and format drift.
- Run a pre-submission similarity check.
- Compute the effective index and revise any unattributed overlap — catches what the first four steps missed.
- Back up the library.
Verification is the last step, never the first. A workflow that begins with a similarity check is editing for a score rather than writing with integrity — the precise inversion this unit exists to prevent.
9. Summary
- Paraphrasing is a cognitive skill requiring comprehension, restructuring and citation — not synonym substitution. The five-step method is the foundation.
- Citation styles are formal languages that vary by discipline. IEEE dominates computing; style consistency is non-negotiable.
- Reference management software eliminates manual citation errors and integrates with Word and LaTeX. Follow the six-step workflow.
- Detection tools produce similarity reports that require human interpretation. An effective index below 10–20 per cent, after excluding references, quotations and methodological phrases, indicates compliance with UPES standards.
10. Exam Preparation
- Memorise the structure of one citation style in detail — preferably IEEE. Produce both an in-text citation and a reference entry from memory.
- Practise paraphrasing three short technical passages using the five-step method.
- Install Mendeley or Zotero before the next assignment and build a library of 5–10 references.
- Submit a sample paragraph through a free checker before formal submission, to identify accidental overlap.
- Review the UPES Academic Integrity Code and the specific similarity threshold for theses in your department.
10.1 Short-Answer Questions
- Differentiate between paraphrasing and patchwriting. Why is the latter ethically problematic? 3 marks
- List the four components of a complete IEEE in-text citation and reference entry. 2 marks
- What are the three databases against which Turnitin compares submissions? 3 marks
- State the UPES-recommended maximum similarity index for a B.Tech thesis, excluding references and quotations. 2 marks
10.2 Medium-Answer Questions
- Explain the five-step paraphrasing method with an example drawn from a CS research paper of your choice. 5 marks
- Compare Mendeley and Zotero as reference management tools. Under what circumstances would you prefer one over the other? 5 marks
- A student receives a Turnitin similarity index of 32%. Describe, step by step, how they should analyse and act upon this report. 5 marks
10.3 Long-Answer and Application
- Application. You are writing the Literature Review chapter of a B.Tech thesis on Privacy-Preserving Machine Learning in Healthcare. Draft (a) two properly paraphrased passages from a selected paper with IEEE citations, (b) the corresponding IEEE reference entries, and (c) a short justification of your paraphrasing choices. 10 marks
- Analytical. Critically evaluate: “A high similarity index from Turnitin always indicates plagiarism.” Discuss the limitations of similarity-detection tools and the importance of human interpretation. 10 marks
- Constructive. Design a personal anti-plagiarism workflow for a B.Tech thesis project, integrating paraphrasing technique, citation style, reference management and similarity checking. Justify each step with reference to the ethical principles discussed in Unit III. 10 marks
11. Suggested Reading
- American Psychological Association (2020). Publication Manual of the American Psychological Association (7th ed.).
- IEEE Editorial Style Manual (2022). IEEE.
- Roig, M. (2015). Avoiding plagiarism, self-plagiarism, and other questionable writing practices: A guide to ethical writing. Office of Research Integrity.
- University of Wisconsin–Madison Writing Center (2023). Paraphrasing.
- UPES (2023). Academic Integrity and Plagiarism Policy. UPES Student Handbook.