ICPS Integrity Check Criteria for Organizers

AI-generated text

Any percentage indicated as AI-generated is potentially problematic. Note that this check may also find sections of papers that were written by, or with the help of, tools such as ChatGPT, and this may be acceptable in some cases if the author declared their use of such tools. The formal ACM policy on this is as follows: “The use of generative AI tools and technologies to create content is permitted but must be fully disclosed in the Work. For example, the authors could include the following statement in the Acknowledgements section of the Work: ChatGPT was utilized to generate sections of this Work, including text, tables, graphs, code, data, citations, etc.”

  • Papers showing fraudulent use of AI-generated text must be rejected. Please report any such instances to ACM.
  • In cases where the legitimacy of AI-generated text is unclear, the author should be asked to explain.

Over-citation

This check is currently configured to identify cases where the same author is cited more than three times and this is more than 10% of the total citations.

  • Papers where the over-citation is clearly fraudulent must be rejected. Please report any such instances to ACM.
  • In cases where the legitimacy of the multiple citations is unclear, the author should be asked to explain and make corrections if necessary.

Tortured phrases

Such phrases (example: “fuzzy logic” is rendered as “fluffy reasoning”) should be assessed for frequency and context. It is possible that the check will identify stray phrases that are actually acceptable, but in general it should be clear when the usage is fraudulent. In general, papers with such problems should be rejected as fraudulent.

  • Papers where the tortured phrases are clearly fraudulent must be rejected. Please report any such instances to ACM.
  • In cases where the legitimacy of the phrases is unclear, the author should be asked to explain and make corrections if necessary.

Similarity score

Papers with similarity scores of 25% or above require further investigation. In some cases where there is a high score, the author may have reused text from a legitimate source such as a preprint of the same paper posted to arXiv, ResearchGate, etc.

  • Papers where there is clear evidence of plagiarism must be rejected. Please report any such instances to ACM.
  • In cases where the legitimacy of the paper is unclear, the author should be asked to explain the high similarity score.

Retracted papers in references

Reference lists should not include retracted articles. This can be a sign of papermill activity.

  • Any papers in which this occurs should be re-examined to check their overall legitimacy, and rejected if necessary.
  • If the inclusion of retracted articles in the reference list appears to be a genuine mistake, the author should be required to remove those references.

Publish with ACM

ACM's prestigious conferences and journals seek top-quality papers in all areas of computing and IT. It is now easier than ever to find the most appropriate venue for your research and publish with ACM.

Publish your work

Lifelong Learning

ACM offers lifelong learning resources including online books and courses from Skillsoft, TechTalks on the hottest topics in computing and IT, and more.

techpacks