About This Project

Tortured phrases, characterized by complex and ambiguous language, obscure authors' real meanings and pose a growing threat to scholarly credibility. For example, "counterfeit neural organizations" instead of "Artificial Neural Networks" and "portrayal learning" instead of "representation learning." To counter this issue, we will analyze 15,000 articles, investigating how the appearance of the tortured phrases affects the readers, publishers, and institutions.

Ask the Scientists

What is the context of this research?

While investigating various research papers, we encountered tortured phrases that challenged our understanding. For example, in an article in 2021, we read the sentence: "The secret segments move dependent upon the disease. Coronary vein disease, stroke, and periphery supply course ailment incorporate atherosclerosis." Upon further examination, we realized the intended statement was: "The underlying mechanisms vary depending on the disease. Coronary artery disease, stroke, and peripheral artery disease involve atherosclerosis." These sentences made us struggle to grasp the author's intended meaning. So, we aim to understand their impact on scholarly credibility and the reputation of publishers and institutions.

What is the significance of this project?

Our project holds significant implications for scholarly research and communication, prioritizing clarity and transparency in scientific literature. We seek to enhance comprehension among peers and the public by addressing the challenge of tortured phrases and complex language. Through meticulous analysis of 15,000 articles, we plan to identify and replace tortured phrases, collaborating with authors, publishers, and institutions for meaningful change. By establishing guidelines and encouraging a cultural shift towards more explicit language, our work aims to transform scholarly communication, elevating accessibility and the impact of scientific knowledge for a lasting effect.

What are the goals of the project?

We will use the fund to pay for server services, training model processes, students, and volunteers. We will follow four specific steps, including gathering articles from ArXiv & Crossref API using Talbots. We then identify & quantify tortured phrases by calculating the ratio of tortured phrases to total pages. Thirdly, identify other quality metrics (Interest Score, Citation, and Altimetric index) to evaluate the preprint quality. Finally, analyze the ratio of tortured phrases to the quality score to check how their quality is affected by the presence or absence of tortured phrases. Computer Science, Engineering, and education are our fields of expertise, so we only focus on them in this research. We will share our findings with the community through publication on Research Square.