ProQ-KG: Integrated Cyber-Physical Production System Knowledge Graph for Quality Issue Analysis

This paper proposes an integrated ontology that unifies PPR (Product-Process-Resource) and FMEA (Failure Mode and Effect Analysis) modeling approaches, and presents ProQ-KG, a knowledge graph-based framework that enables multi-stakeholder collaboration in analyzing root causes of quality issues within complex cyber-physical production systems, demonstrated through real-world use cases from the automotive industry.

This paper presents ProQ-KG, a knowledge graph framework that integrates PPR (Product-Process-Resource) models with FMEA (Failure Mode and Effect Analysis) through a unified ontology. Building upon prior work, the authors contribute a formally defined set of six dependency link types, an Investigation State Marker (ISM) mechanism, and a collaborative quality analysis lifecycle. Compared to existing FMEA-based ontology approaches, this work demonstrates clear advances in cross-disciplinary knowledge integration, multi-view coordination, and support for knowledge evolution. The originality is high.

Strengths:
Use of real industrial data enhances external validity
The methodology is systematically designed, with formal definitions of dependency links and a clear seven-step lifecycle
The use of W3C standards (RDF/OWL/SPARQL) ensures interoperability and extensibility

Weaknesses and Suggestions:

1 Insufficient coverage of recent related work: In the "Ontological/Knowledge Graph Based Quality Issue Analysis" subsection, although several KG-related studies are cited, recent advances are not adequately covered. Important references are missing, such as KGroot: A knowledge graph-enhanced method for root cause analysis and A comprehensive survey on root cause analysis in (micro) services: Methodologies, challenges, and trends.

2 In Section 6, the paper presents the constructed knowledge graph but does not report quantitative metrics, such as query execution time, accuracy of root cause identification, or user interaction efficiency.

3 The evaluation does not include a comparison with traditional FMEA methods or alternative KG-based approaches, making it difficult to assess the relative advantages of the proposed method.

4 Although the paper addresses knowledge gaps among multiple stakeholders (e.g., quality managers, process experts), no systematic user study or usability evaluation is reported to validate the effectiveness of the proposed approach in supporting cross-domain collaboration.

5 As noted in the data artifact assessment, the provided resources lack complete SPARQL query sets and full construction scripts, which limits the replicability of the experiments.

6 The manuscript is well-structured, clearly written, and technically precise. The logical flow from system requirements, through methodology and implementation, to use case evaluation is coherent and easy to follow. Figures and tables (e.g., the ontology overview in Figure 4, the implementation pipeline in Figure 5, and the KG visualization in Figure 6) effectively support the narrative. The overall writing quality is high.

However, there are several placeholders (indicated by "??") that need to be corrected throughout the manuscript, including but not limited to:

"Table 6 and ?? (see Appendix) shows an excerpt overview..."

"We used our prototype pipeline discussed in ?? to process the data..."

To accommodate the dependency links introduced in ?? , we developed several object properties (OP) that link between classes in ProQ ontology as follows:

We used our prototype pipeline discussed in ?? to process the data, construct the knowledge graph and perform analysis.

"We summary our contributions with respect to our identified requirements (cf. ?? ) as follows:"

1.Originality
This paper presents "ProQ-KG", a knowledge graph-based framework designed to integrate Product, Process, Resource (PPR) models with Failure Mode and Effect Analysis (FMEA) in Cyber-Physical Production Systems (CPPSs). The authors propose the ProQ ontology to harmonize heterogeneous domain knowledge and introduce a 7-step collaborative quality issue analysis lifecycle utilizing "Investigation State Markers". The framework is implemented using semantic web technologies (RDF/OWL, SPARQL) and three use cases in a real-world laser beam welding scenario in the automotive industry.
2.Significance of the Results
Lack of Quantitative Evaluation: The current evaluation (Section 6) is largely descriptive and centered on three use cases and example SPARQL queries. This is sufficient to show that the approach can be implemented and exercised in one industrial setting, but it does not yet substantiate stronger claims regarding efficiency, effectiveness, improved collaboration, or better root-cause identification. The paper itself later acknowledges that future work should include quantitative metrics such as query execution time, accuracy of root-cause identification, resolution effectiveness, and user interaction efficiency. This confirms that the current evaluation remains preliminary.
Clarification of Novelty vs. Prior Work: The paper heavily relies on the authors' previous work on the FPI model (Reference [8]). It is currently somewhat blurred where the FPI conceptual model ends and the ProQ-KG contribution begins.
Ontology Evaluation: Since the ontology is central to the paper’s contribution, it would be important to evaluate it more formally. At present, the development process is described, but the paper does not report consistency checking, competency-question-based validation, SHACL constraints, or DL-based reasoning tests. The discussion section explicitly acknowledges that formal ontology evaluation is still future work.
Underdeveloped Marker Semantics: While the ISM concept is introduced, its formal semantics (e.g., in OWL) are not elaborated. How markers interact with reasoning or propagate through dependencies is only described procedurally.
Limited Comparison with Alternatives: While Table 1 positions the work against prior approaches, the paper would be strengthened by a more explicit comparative discussion of what ProQ-KG enables beyond existing graph-ontology-based solutions, even if a full benchmark is not feasible.
3.Quality of Writing
Writing: The term “Failure Mode and Effect Analysis (FMEA)” is inconsistently introduced. It is first referred to as “potential failure modes and effects (FMEA)” and later redefined as “Failure Mode and Effect Analysis (FMEA)”. Please ensure consistent and correct terminology throughout the paper.
Internal Inconsistency in the Lifecycle Description: The paper introduce a 7-step collaborative quality issue analysis lifecycle based on "Investigation State Markers", but the actual list only describes step 1 through Step 5. This should be corrected.
Broken Cross-reference: There are visible “??” placeholders in multiple places.
Figure Readability: Figure 4 is very cluttered and the text is extremely hard to read even when zoomed in. Please redesign this figure or split it to make the core classes and object properties legible.
Table 4 Anomalies: Table 4 appears inconsistent: the total number of object properties does not match the per-module counts.
Figure 6: Figure 6 is difficult to interpret and currently offers limited analytical value beyond indicating graph scale/connectivity.
Definition: The definitions of PPR-to-FMEA and PPR-to-PPR dependencies contain several inconsistencies. In Definition 6, the description incorrectly refers to FMEA, which seems to be a copy-paste error.
Overall, the paper presents an interesting and relevant framework for integrating PPR and FMEA using knowledge graphs. However, the current version suffers from limited quantitative evaluation, incomplete ontology validation, and several writing and consistency issues. Addressing these concerns would significantly strengthen the paper.