Citation Detective: Audit a Literature Review
Citation Detective: Audit a Literature Review
A draft literature review cites real papers. Check every claim against its source, catch the retracted papers and fix the broken references.
Briefing
A student has sent you a draft literature review on machine learning for malware detection. Every source it cites is real, and every one can be checked online.
Some claims are faithful to their sources. Some twist what the source actually says. Two cite papers that have since been retracted, and the reference list has errors a careful reviewer must catch.
Your job is the supervisor's job: open each source, compare it with the claim, look up the records in Crossref and the Retraction Watch Database, and correct the references. Reference managers format citations; they do not check them.
Topics you practise
Evidence
draft-literature-review.txt
The draft section with 8 numbered claims and 7 references. Every reference has a link or DOI you can follow. (3 KB)
Source: Written for this lab; the cited papers and their records are real (arXiv, Crossref, Retraction Watch). Draft text by EpochZero; each paper belongs to its authors.
Suggested tools: arXiv (read the abstracts), Crossref search (search.crossref.org), the Retraction Watch Database (retractiondatabase.org), doi.org, the APA 7 and IEEE style guides.
Which claims are fully supported by the source they cite? Give the claim numbers.
Format: numbers in ascending order, like 2,6,8.
Claim 2 reverses a result from EMBER. According to its abstract, which model performed better?
Format: the model name.
Claim 3 says batch normalisation was essential for training MalConv. What does the MalConv abstract actually say it did?
Format: one word: helped or hindered.
Claim 6 says the pitfalls Arp et al. found are rare. Which single word in their abstract describes how common the pitfalls are?
Format: one word.
Which references have been retracted? Give the reference numbers.
Format: numbers in ascending order, like 1,3.
Enter the DOI of the retraction notice for reference [7].
Format: a DOI, like 10.xxxx/xxxxx.
In which year was reference [6] retracted?
Format: a year.
Reference [1] gives the wrong year. What is the correct year of the EMBER preprint?
Format: a year.
Reference [5] lists the authors in the wrong order. Who is really the first author? Give the surname.
Format: a surname.
Reference [3] has the wrong page range. What is the correct range?
Format: first-last, like 10-25.
APA 7 has a rule for how a DOI is written. How should the DOI of reference [3] appear in an APA 7 reference?
Format: the DOI exactly as APA 7 writes it.
If the draft switched to IEEE style, how would the authors of reference [1] be written?
Format: the author list as IEEE writes it.
Reference [4] cites the arXiv preprint, but the paper was later published after peer review. At which conference?
Format: the conference name.
Which of the cited works release a dataset that other researchers can reuse? Give the reference numbers.
Format: numbers in ascending order, like 2,4.
Each task is worth 100 points: minus 15 per wrong answer and 25 for a hint (never below 40). Finish when you are done; 14 unsolved tasks score 0. Your progress is saved on this device.