Literature Review Data for Neuro-Symbolic Sequence Generation in SME Process Planning

This dataset contains a comprehensive collection of literature and evaluation data in support of the study, 'Neuro-Symbolic Sequence Generation for Automated Process Planning in SMEs'. The data provides an empirical basis for the systematic evaluation of AI technologies and the subsequent derivation of a modular, neuro-symbolic AI pipeline pattern designed for automated macro-routing planning in SMEs. Within the associated study, this architecture is synthesized using the Architecture Tradeoff Analysis Method (ATAM) to explicitly balance critical non-functional engineering requirements. Content of this dataset: Literature Source List: A curated list of scientific publications referenced in the paper, covering multimodal order analysis (e.g., CNN, ViT, GNN, OCR), autoregressive sequence generation (e.g., RNN, Generative Transformers, CVAE, Graph Decoders), and symbolic knowledge infusion strategies (Early, Intermediate, and Late Fusion). Qualitative Technology Assessment: A comparative ranking of the identified technologies based on SME-specific non-functional requirements, mapped along an ordinal suitability scale. Detailed Technology Justification Matrix: A granular assessment and scientific reasoning for the scoring of specific AI architectures and fusion methodologies against the primary evaluation parameters: Process Sequencing & Validity of Results: The architectural capability to capture routing dependencies, enforce strict manufacturing constraints, and prevent stochastic hallucinations. Small Data (Practicability): Performance, sample efficiency, and training stability in strict "small data" environments (N≤1000). Limited Resources (Practicability): Suitability for lightweight deployment and inference on local, resource-constrained industrial workstations. Adaptability & Simplicity: The system's structural flexibility regarding physical shop-floor changes and the minimization of specialized machine learning maintenance efforts. Usage Note: This dataset is intended for researchers and engineers in the fields of Computer-Aided Process Planning (CAPP), Neuro-Symbolic AI, and Industrial Automation who wish to replicate the technology selection process or adapt the synthesized framework to other manufacturing domains.