Textual analysis of artificial intelligence manuscripts reveals features\n associated with peer review outcome

Abstract

We analysed a dataset of scientific manuscripts that were submitted to\nvarious conferences in artificial intelligence. We performed a combination of\nsemantic, lexical and psycholinguistic analyses of the full text of the\nmanuscripts and compared them with the outcome of the peer review process. We\nfound that accepted manuscripts scored lower than rejected manuscripts on two\nindicators of readability, and that they also used more scientific and\nartificial intelligence jargon. We also found that accepted manuscripts were\nwritten with words that are less frequent, that are acquired at an older age,\nand that are more abstract than rejected manuscripts. The analysis of\nreferences included in the manuscripts revealed that the subset of accepted\nsubmissions were more likely to cite the same publications. This finding was\nechoed by pairwise comparisons of the word content of the manuscripts (i.e. an\nindicator or semantic similarity), which were more similar in the subset of\naccepted manuscripts. Finally, we predicted the peer review outcome of\nmanuscripts with their word content, with words related to machine learning and\nneural networks positively related with acceptance, whereas words related to\nlogic, symbolic processing and knowledge-based systems negatively related with\nacceptance.\n

Ähnliche Arbeiten