In this post, Kurt Spurlock, Quality Manager and Brandon Jernigan, UX Designer & Writer (AJE & Research Square) share their insight into how both publishers and authors can benefit from Artificial Intelligence-based services that are designed to make manuscript preparation as easy as possible.
Publishers and authors could benefit from standardized formatting with the support of automation.
Authors often complain about the tedious work of formatting. Imagine hearing the disappointing news that your paper has been rejected by a journal, only to discover that your alternative choice for publication has an entirely different style preference, which will require hours of work before the article can even be considered.
Journal-specific style guides are meant to make the publishing process more consistent and ultimately to reduce the time and labor involved in publication. But the plethora of expectations across journals and the fact that papers are often rejected outright for minor formatting differences means that authors, journals, and the research community in general experience unnecessary delays in sharing valuable discoveries.
In other words, using AI to apply a standard style is not simply a way of streamlining production work, but it could also reduce serious obstacles in the initial submission and review process.
To make sure our authors do not have to waste time overcoming this mostly unnecessary hurdle to entry, all Hindawi journals operate a “format free” submission process that means authors can provide their manuscript without the need to format or structure it in a specific way. Our internal quality control teams carry out required formatting work if and when a manuscript manuscript makes it past peer review. This means authors can spend their time on the important work of conveying and describing their science, and not on meeting the idiosyncratic requirements of a journal’s style guide.
Triaging for multiple issues
AI can handle decisions for large amounts of routine information while simultaneously identifying irregularities or nuanced cases for human attention.
AI technology can reduce the signal-to-noise ratio and chronic stress of human decision making in an overburdened process like publishing.
Both journals and authors could benefit from a submission process that automatically triages papers and identifies potential issues. For example, a paper may need language editing before review, or important information missing from tables that need, for example, language editing would be a benefit to authors and publishers.
These problems are not always straightforward or easy for NLP or ML tools, especially when it comes to data preparation, but AI has become a reliable method for dealing with large amounts of routine information while simultaneously identifying irregularities or nuanced cases for human attention.
In addition to simplifying manuscript preparation and journal submission, machine learning has the potential to support ethical oversight in academic publishing in ways that aren’t possible or feasible with human oversight alone. Simple but important mistakes, such as poor image resolution or self plagiarism, could be identified as part of a triage process with more accuracy and less human effort.
AI could screen manuscripts for missing information or routine issues with ethics statements, patient consent, methods, and so on. Again, although we can imagine the horror of rogue decision-making algorithms, the most effective implementations would create a supportive process that allows authors to resolve issues quickly and editors to focus on issues that require their full attention. In fact, identifying the most insidious cases of academic fraud, especially those involving data or image manipulation, will likely require the combined effort of powerful AI tools and human expertise.
Spelling and grammar tools can learn the language of academic writing.
Most writers and editors are accustomed to the grammar and spelling tools built into word processors like Microsoft Word and Google Docs. These tools depend on natural language processing and have become increasingly sophisticated through machine learning and large text corpora. As in the examples above, they help increase the consistency of language, grammar, usage, and punctuation while also flagging more complex potential issues for users. Consistency on these issues can also help reviewers focus on the content of papers with fewer stylistic or grammatical distractions (e.g., use of abbreviation and acronyms in a paper).
In fact, the history of spell check tools seems apt. In the 1990s, spell checkers rapidly evolved from simple style checkers to popular multipurpose tools that could parse and analyze text using natural language processing. They show us how accustomed writers and editors can become to using fairly complex technology in a short span of time.
Machine learning has significantly advanced these spelling and grammar tools, but it requires vast amounts of high-quality data. With advancements in AI and data preparation, authors and publishers are likely to benefit from more customizable language tools that can consider and incorporate the particular conventions of academic writing in general as well as the preferences, terminology, and expectations of individual disciplines or journals.
This post was co-written by Kurt Spurlock, Quality Manager and Brandon Jernigan, UX Designer & Writer (AJE & Research Square). It is distributed under the Creative Commons Attribution License (CC-BY). The illustration is by Hindawi and is also CC-BY.