Intelligent Publishing Workflow Redefines China’s Book Production
In an era defined by digital acceleration, artificial intelligence (AI), and evolving reader behaviors, the traditional publishing industry in China is undergoing a profound transformation. A groundbreaking study published in China Media Technology has introduced a new paradigm for book production—intelligent publishing workflow—that promises to elevate quality, streamline efficiency, and align with national goals of building a culturally strong nation by 2035.
Led by Liu Ruidong from Dongbei University of Finance and Economics Press, in collaboration with Liu Changming, Gao Guolian from Beijing Founder Electronics Co., Ltd., and Gong Xiaohui, also from the university press, the research presents a comprehensive rethinking and redesign of publishing processes through what they term the “intelligent publishing business process model.” This innovation emerges not as a mere technological upgrade but as a strategic response to systemic challenges exposed during the pandemic and amplified by rapid digital disruption.
The paper argues that enhancing publishing productivity is central to achieving a “publishing power” status—a key milestone on the path to becoming a socialist cultural power. The authors define a publishing power by four core attributes: strong political guidance, high-quality content, robust production capacity, and effective dissemination. Among these, production capability stands out as both foundational and most vulnerable to inefficiencies rooted in outdated workflows.
For decades, Chinese publishing relied on linear, siloed processes where manuscript creation, editing, typesetting, proofreading, and printing occurred across disconnected platforms. Authors submitted Word documents; editors marked up physical or digital copies; designers used specialized layout software; and proofreaders manually compared versions. Each transition between stages introduced delays, errors, and data loss. When the pandemic struck in 2020, lockdowns disrupted physical workflows, exposing critical weaknesses in operational continuity and remote collaboration.
This crisis catalyzed a shift toward digital transformation. However, many publishers adopted piecemeal solutions—digitizing manuscripts without integrating systems or automating tasks without rethinking roles. The result was often digitized inefficiency rather than true innovation.
The team’s response goes beyond incremental change. Drawing on principles of business process reengineering, they propose a fundamental reimagining of how books are made—from initial submission to final output. Their approach rests on six guiding considerations for modern publishing: unwavering commitment to quality across content, editing, design, and print; strict adherence to regulatory standards such as the “three reviews, three proofreads, one final read” system; integration of innovative, coordinated, green, open, and shared development concepts; adaptation to media convergence; human-centered design that enhances editorial well-being; and responsiveness to technological shifts like AI, big data, and mobile reading habits.
At the heart of their framework lies the concept of “three integrations and one platform”—a unifying architecture designed to overcome fragmentation. First is the deep integration of human editors with artificial intelligence. Rather than replacing people, AI acts as a force multiplier, handling repetitive, rule-based tasks so professionals can focus on higher-value judgment work. Second is the unification of technical platforms used across writing, editing, typesetting, proofreading, and structuring—eliminating costly format conversions and version mismatches. Third is the cross-functional integration of editorial roles, reflecting evolving responsibilities under digital conditions.
All three integrations converge on a single intelligent cloud publishing platform developed through joint experimentation at the Dongbei University of Finance and Economics Press’s editing lab and supported by Founder Electronics’ technology team. This cloud-based environment enables real-time collaboration regardless of location—an essential feature in post-pandemic operations—and supports full lifecycle management of content assets.
One of the model’s most significant contributions is its redefinition of the manuscript itself. In legacy systems, the term “original manuscript” typically refers to the author’s final draft after editing. But in this new workflow, the starting point shifts to what the authors call the “final-formatted original manuscript.”
This refined document results from several pre-processing steps: automated AI review detects political sensitivities, spelling errors, punctuation inconsistencies, and duplicate content; technical editors apply standardized templates for consistent formatting; and the revised version undergoes preliminary review by planning editors before being sent back to the author for confirmation. Only then does the responsible editor begin substantive editing.
By ensuring the manuscript meets the industry standard of “complete, clear, and finalized” (qi qing ding) before entering the main editorial phase, the model drastically reduces downstream corrections and revision cycles. Errors caught early prevent cascading issues later in production, saving time and preserving quality.
Crucially, the entire workflow remains fully compliant with China’s stringent publishing regulations. The “three reviews, three proofreads, one final read” requirement is preserved—not circumvented—through intelligent augmentation. Manuscripts still pass through four formal review stages: initial assessment by a planning editor, substantive editing by the responsible editor, secondary review, and final approval. Each role maintains distinct accountability per national guidelines.
Proofreading follows a hybrid human-machine protocol aligned with recommendations from the China Publishers Association, which recognizes computer-assisted proofreading as capable of fulfilling one complete proofreading round when properly implemented. In practice, AI performs mechanical checks—detecting typos, formatting deviations, and structural anomalies—while human proofreaders concentrate on semantic accuracy, logic, and context-sensitive judgments.
A dedicated technical proofreader handles file integrity and consistency, while a final reading ensures no last-minute errors slip through prior to printing. Quality inspection remains a mandatory step conducted independently. Thus, even as automation increases, oversight does not diminish—it evolves.
Artificial intelligence serves as the engine driving performance gains. Three primary functions illustrate its impact: intelligent proofreading, intelligent comparison, and intelligent typesetting.
Intelligent proofreading leverages natural language processing and trained models to identify linguistic inaccuracies, including incorrect usage of politically sensitive terms, improper punctuation, inconsistent numerals, and unit discrepancies. It also flags potential plagiarism and redundant passages within or across chapters. Because it operates continuously throughout the workflow, feedback loops shorten dramatically.
Intelligent comparison replaces labor-intensive manual comparison methods such as folding pages or red-marking changes. By analyzing two versions of a text down to the character level, the system instantly highlights additions, deletions, and modifications. Editors and proofreaders gain immediate visibility into revisions, reducing the risk of overlooked edits and accelerating turnaround times.
Intelligent typesetting uses predefined style templates to automatically format content according to genre-specific rules—academic monographs, textbooks, trade books, etc. Currently, over 60% of layout work can be completed without manual intervention, freeing designers for complex visual treatments and cover development. More importantly, template-driven consistency ensures brand coherence across titles and series.
Together, these capabilities do more than improve speed—they enhance reliability. Human fatigue, especially during repetitive tasks, leads to variable performance. Machines, once calibrated, deliver uniform precision. This allows publishers to maintain high-quality outputs even under tight deadlines or increased volume.
Perhaps the most transformative aspect of the model is its effect on editorial roles. Far from displacing workers, AI enables a strategic repositioning of human talent. The authors describe this as “reduction and empowerment” (jian yu fu).
“Reduction” refers to offloading routine cognitive burdens—the constant vigilance required for grammar, syntax, citation styles, and typographical correctness. These low-level concerns consume substantial mental energy, leaving less bandwidth for deeper analytical thinking. With AI managing them, editors can redirect attention to substantive issues: factual accuracy, logical coherence, disciplinary rigor, and ideological alignment.
“Empowerment” means expanding the scope and value of editorial work. As support staff evolve into technical editors, they assume responsibility not just for typesetting but for enforcing design standards and identifying layout flaws. Text editors deepen their expertise to become subject-matter specialists—academic or discipline-focused editors—who engage critically with content rather than merely polishing prose.
Planning editors, meanwhile, broaden their remit to include market analysis and audience engagement strategies, transforming into project editors who oversee both intellectual and commercial dimensions of publication. This evolution reflects a broader trend in knowledge industries: the fusion of creative, technical, and managerial competencies within individual roles.
Moreover, the emotional and psychological dimensions of editorial labor cannot be ignored. Publishing has long faced challenges related to job satisfaction, burnout, and professional recognition. By alleviating drudgery and elevating the intellectual stature of daily tasks, the intelligent workflow contributes to greater occupational fulfillment. Editors report feeling more respected, more impactful, and more secure in their positions—key indicators of sustainable workforce development.
From a technological standpoint, the shift to a unified cloud platform marks a decisive break from analog-era practices. All activities occur within a centralized digital workspace accessible anytime, anywhere. This facilitates seamless handoffs between team members, transparent tracking of progress, and version control with full audit trails.
The platform natively generates multiple output formats—EPUB for e-books, PDF for print-ready files, XML for database integration, and Word for external sharing—enabling multi-channel distribution without rework. Furthermore, every action leaves a data footprint: how long each review took, how many errors were corrected, which sections underwent the most revisions.
These process data, when aggregated and analyzed, offer unprecedented insights into workflow bottlenecks, individual performance patterns, and quality trends. Publishers can use this information to refine training programs, optimize resource allocation, and predict delivery timelines with greater accuracy. Over time, machine learning algorithms could suggest optimal workflows based on historical success rates.
Looking ahead, the researchers emphasize that digital reinvention is not a one-time event but an ongoing journey. They identify three future directions shaped by emerging technologies.
First is collective intelligence (qunti zhinen), inspired by initiatives like Wikipedia, open-source software development, and crowdsourced Q&A platforms. As outlined in China’s New Generation Artificial Intelligence Development Plan, collective intelligence harnesses networked human participation coordinated by AI systems to solve complex problems collaboratively. Applied to publishing, this could mean distributed peer review networks, community-driven fact-checking, or co-creation models involving readers in content development.
Second is structured data editing, moving away from flat text toward semantically rich, machine-readable content. Instead of treating a book as a sequence of paragraphs, future workflows may decompose it into discrete knowledge units—concepts, definitions, arguments, evidence—tagged with metadata and linked logically. This enables dynamic repurposing: generating summaries, extracting teaching modules, feeding into recommendation engines, or powering AI tutors.
Third is knowledge graph construction, where interconnected nodes represent entities (people, places, theories, events) and relationships (causality, influence, contradiction). A textbook on economics, for example, could generate a living knowledge map showing how Keynesian theory relates to fiscal policy debates, illustrated with real-world case studies and updated with new research findings. Such graphs transform static publications into adaptive learning environments.
To realize this vision, publishers must move beyond viewing technology as a tool and embrace it as a partner in value creation. That requires investment in infrastructure, talent development, and organizational culture. Legacy mindsets that equate quality with manual effort must give way to a new ethos where intelligence—both human and artificial—is leveraged strategically.
It also demands policy support. While the central government promotes cultural strength and technological self-reliance, specific incentives for publishing innovation remain limited. Greater public-private collaboration, funding for R&D pilots, and clearer regulatory frameworks for AI-generated content would accelerate adoption.
Nonetheless, early adopters are already seeing results. Pilot implementations at Dongbei University of Finance and Economics Press have demonstrated measurable improvements: reduced production cycles by up to 40%, lower error rates, higher employee satisfaction, and expanded digital product offerings. Other academic and educational publishers are beginning to follow suit.
Critics may argue that automation risks homogenization or diminishes editorial autonomy. Yet the evidence suggests otherwise. Standardization applies only to mechanics, not creativity. Editors retain full authority over content decisions—their role becomes sharper, not weaker. And diversity flourishes when resources previously tied up in routine tasks are redirected toward discovering new voices and exploring niche subjects.
Ultimately, the intelligent publishing model represents more than a technical upgrade. It embodies a philosophical recalibration—one that honors tradition while embracing progress, balances control with agility, and places human dignity at the center of industrial transformation.
As China strives to project soft power globally and cultivate an informed citizenry domestically, the integrity and reach of its publishing sector matter profoundly. By rebuilding workflows around intelligence, integration, and inclusivity, the industry can fulfill its mission with renewed purpose and precision.
The road ahead will require continuous learning, ethical vigilance, and collaborative spirit. But if the insights from this research are any indication, the future of publishing is not only digital—it is deeply human.
Liu Ruidong, Liu Changming, Gao Guolian, Gong Xiaohui. Intelligent Publishing Workflow Redefines China’s Book Production. China Media Technology. DOI: 10.19483/j.cnki.11-4653/n.2021.07.005