Single-Source Publishing Workflow of the journal Philosophy and the Mind Sciences

Institute for Philosophy II
Ruhr University Bochum
44780 Bochum
Germany

We aim to provide an efficient workflow that enables one to run a journal independently and with minimal financial overhead (while relying on volunteers from the community, i.e., the editorial team).

Description

The following description presents the current workflow (as of June 2026) of the journal Philosophy and the Mind Sciences (PhiMiSci). PhiMiSci is an interdisciplinary, peer-reviewed, diamond open access journal indexed in SCOPUS (https://philosophymindscience.org/).

Our single-source publishing workflow accepts DOCX (using the Zotero plugin for citations) and ODT as the primary submission formats. These files are converted into Markdown, which serves as the authoritative and central file for all further processing and the generation of individual galleys (e.g., PDF and HTML). Articles are submitted, reviewed, and published via OJS, while typesetting is handled externally using a web application called Magic Manuscript Maker (MMM), which was mainly developed by Thomas Jurczyk and Felix Kopecky as a contributor (particularly of the AI Bibliographer). The MMM bundles the various processing steps into a single application, making it decentralized (eliminating the need to install dependencies on each editor’s computer) and enabling the organization and sharing of article projects and files among editors.

Citations are automatically processed: either via Pandoc, if citations have been inserted using the Zotero plugin; or with a script developed by us if citations have been inserted manually; the script also requires providing reference lists in BibTeX format (which we currently generate using Claude Sonnet 4.6 on the basis of the plain-text reference lists). Since the script does not usually recognise all manually inserted citations, this process is not fully automated.

Formats

• Accepted input formats: DOCX, ODT, MD
• Generated output formats: PDF, HTML (JATS XML in preparation)
• Single-Source of Truth format: Markdown
• Intermediary formats: BibTeX (for references), YAML (for metadata), TeX (for conversion to PDF)

Software

• OJS: Used for submission, peer review, and publication of articles.
• Pandoc: Handles the core document conversion tasks (called via the Magic Manuscript Maker).
• LaTeX: Used for PDF generation (called via the Magic Manuscript Maker).
• Docker: Runs the Magic Manuscript Maker and its services on a server.
• Claude Sonnet 4.6: Currently used as the base model for converting reference lists to BibTeX (experimental; future plans include using KI:connect or local models).
• Magic Manuscript Maker (MMM): A wrapper application for the typesetting process, including user and project management.

Staffing requirements

• Number of people involved:
  • Submission handling (peer review organization, editorial decisions): ~27 editors
  • Typesetting: 2–4 editors (planned: training additional editors to handle typesetting)
• Required skills/training:
  • Basic knowledge of Markdown and the workflow
  • Familiarity with file uploads and the correct sequence of actions in the system
  • For edge cases and troubleshooting: basic LaTeX knowledge and advanced understanding of Markdown and Pandoc

Extra requirements imposed on authors

Authors are asked to:

• Check and update article metadata in OJS.
• Insert citations using Zotero or adhere to APA7 citation format.
• Apply basic Markdown markup for section titles, internal links, and figures.

Costs

• Hosting the Magic Manuscript Maker: ~€45/month (costs could be reduced with a less powerful server).
• Claude Sonnet API: ~0.50 €/month (average).

Challenges and Trade-Offs

The central trade-off is between having a customisable workflow that can accommodate many exceptions, and having a simple workflow that can be used with little training. The web application makes the workflow relatively simple; however, if articles have not been prepared carefully (e.g., missing reference data) or if there are special cases (e.g., complex tables or formulae), knowledge of the markup formats that we use (in particular, Markdown and LaTeX) is required to handle these cases.

Overall, one challenge is therefore that typesetting still takes a considerable amount of time, especially when citations have been added manually by the authors and our script does not automatically recognise all of them. Another challenge is that someone needs to maintain, fix, and develop the core systems (MMM; TeX-class).

For new typesetters, the workflow can be a bit time consuming and even frustrating at the beginning, especially if they are not used to MD / OJS. At the end, however, the web application makes the workflow very rewarding in producing high class PDF (and HTML) with a layout exactly as you want it with a click of a button. This also makes it possible to typeset between 30-50 submissions per year with very few people.

Article metadata

The article metadata currently includes the following fields:

title: Main title of the article.

subtitle: Subtitle (if applicable).

author name(s): Full name.

affiliation(s): Institutions.

email(s): Email address.

orcid(s): ORCID ID.

keywords: List of keywords.

abstract: Summary of the article.

date: Publication year (e.g., 2026).

volume: Volume number (e.g., 7).

issue: Issue number.

artid: Article ID (e.g., 007).

contact email: Primary contact email.

contact name: Primary contact name.