A Deep Multimodal Investigation To Determine the Appropriateness of Scholarly Submissions

Abstract

Present day peer review is a time-consuming process and is still the only gatekeeper of scientific knowledge and wisdom. However, the rapid increase in research article submissions these days across different fields is posing significant challenges to the current system. Hence the incorporation of Artificial Intelligence (AI) techniques to better streamline the existing peer review system is an immediate need in this age of rapid scientific progress. Among many, one particular challenge these days is that the journal editors and conference program chairs are overwhelmed with the ever-increasing rise in article submissions. Studies show that a lot many submissions are not well-informed and do not fit within the scope of the intended journal or conference. Here in this work, we embark on to investigate how an AI could assist the editors and program chairs in identifying potential out-of-scope submissions based on the past accepted papers of the particular journal or conference. We design a multimodal deep neural architecture and investigate the role of every possible channel of information in a research article (full-text, bibliography, images) to determine its appropriateness to the concerned venue. Our approach does not involve any handcrafted features, solely depends on the past accepting activity of the venue, and thereby achieves significant performance on two real-life datasets. Our findings suggest that a system of this kind is possible and with reasonable accuracy could assist the editors/chairs in flagging out inappropriate submissions.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 11 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.