Reviewer has chosen not to be Anonymous

Overall Impression: Good

Content:
Technical Quality of the paper: Good
Originality of the paper: Yes, but limited
Adequacy of the bibliography: Yes

Presentation:
Adequacy of the abstract: Yes
Introduction: background and motivation: Good
Organization of the paper: Satisfactory
Level of English: Satisfactory
Overall presentation: Good

Detailed Comments:

Well written paper which represents both a survey and position paper about NeSy within the traffic domain. The topic is very relevant for the inaugural issue of the journal.

I miss additional information about the type of formal knowledge (graph) that is used within the state-of-the-art and how it is represented. The paper mentions different symbolic representations: knowledge graph, ontologies, semantic network, knowledge bases, rules, common sense rules, soft constraints; but there is not a clear definition of those. KGs and rules can come in different flavours depending on the community. Which is the position of the authors with respect to these terms? How formal should the KGs be in a NeSy System? Are they equivalent to ontologies and knowledge bases? Are rules typically propositional rules, datalog rules, etc.

Similarly, in the inference section I miss some discussion about inference types and engines using KGs. I believe the work described here is very relevant in this line where they use (a fragment of) OWL/Description Logics + datalog rules:
https://www.oxfordsemantic.tech/blog/reasonable-vehicles-rule-the-road

Section 4.2 also mentions symbolic reasoning, but the symbolic representation seems to be reduced to rules.

Minor:
- I believe NeSy is key for explainability, human readable may not be equivalent to human understandable (same for explainability of symbolic systems). NeSy will need to be combined with HCI techniques.
- This knowledge is parsed into a web standard -> do you mean represented via W3C standards like OWL?

Reviewer has chosen not to be Anonymous

Overall Impression: Good

Content:
Technical Quality of the paper: Good
Originality of the paper: Yes, but limited
Adequacy of the bibliography: Yes, but see detailed comments

Presentation:
Adequacy of the abstract: Yes
Introduction: background and motivation: Limited
Organization of the paper: Needs improvement
Level of English: Satisfactory
Overall presentation: Good

Detailed Comments:

The key contributions of the paper are:
1. building a technical task taxonomy in the Traffic Domain including Safety, Perception, and Inference;
2. describing the current state of research in namely safety, perception, and inference, and the research requirements for neural-symbolic AI system in traffic domain.
3. discussing the significance of combining symbolic and neural methods for robust and interpretable traffic understanding, along with some related work.
4. presenting three research directions: representative evaluations for traffic systems, comprehensive knowledge for traffic AI models, and NeSy user-facing models.

The main problems are:
1. traffic domain is a vast research field, is the traffic task classification given in this paper a adequate method？Maybe Figure 1 is the tasks related to traffic understanding instead of traffic domain. in addition, the tasks under the classification are not consistent in granularity. for example, dynamic driven tests vs event detection.
2. The paper presents relevant research work paper by paper; it is lack of the systematic analysis and summary from technique perspectives.
3. Section 3 and 4 focus on the requirements and related research in the traffic domain for neural-symbolic reasoning systems. It is better to provide the possible framework for neural-symbolic AI system in the traffic domain, showcase the components which can integrating symbolic and neural methods in a systematic manner.
4. In three research directions, what is NeSy user-facing models? the description here is too general.

Reviewer has chosen not to be Anonymous

Overall Impression: Good

Content:
Technical Quality of the paper: Average
Originality of the paper: No
Adequacy of the bibliography: Yes

Presentation:
Adequacy of the abstract: Yes
Introduction: background and motivation: Limited
Organization of the paper: Satisfactory
Level of English: Satisfactory
Overall presentation: Average

Detailed Comments:

This paper explores the potential of Neuro-Symbolic AI in the traffic domain. The authors discuss the opportunities of using NeSy reasoning for tasks such as safety, perception, and complex inference. They review prior methods that contribute to this goal in two complementary aspects: robustness and explainability.
The paper also provides background on common representation and aggregation methods in the form of knowledge graphs to explore the synthesis of NeSy approaches and their potential application in traffic domain challenges. The authors suggest that by incorporating richer knowledge into the modeling process, we can unlock new opportunities for the development of intelligent systems capable of reasoning across different modalities and delivering more comprehensive and accurate results. However, they also acknowledge that there are open-ended research directions that require significant innovation, both in the traffic domain and in multimodal reasoning in general.
Overall, the paper provides a general overview of the potential of NeSy reasoning in the traffic domain and highlights the need for further research in this area. To enhance the paper's practicality and applicability, it is recommended that the authors take into account the following suggestions:

- The paper takes a strong theoretical stance, but adding real-world examples would greatly enhance its practical relevance and engagement. Even as a position paper, it remains crucial to present compelling (real-world) examples that emphasize the approach's significance.

- The paper focuses heavily on the benefits of NeSy reasoning, with less focus on the potential limitations or drawbacks of this approach. A more nuanced discussion of the trade-offs between NeSy and other AI approaches would have been helpful. For instance, it would be beneficial to explore how the proposed NeSy approach ensures completeness in representing all rules and constraints, thereby ensuring reliability, robustness, and stability in the solution.

- Furthermore, the paper does not address the potential challenges of integrating symbolic and neural approaches, such as the difficulty of reconciling inconsistencies between the two types of models. Without a clear understanding of how these challenges will be addressed, it is difficult to assess the feasibility and effectiveness of the proposed NeSy approach.

To fully evaluate the potential of NeSy reasoning in the traffic domain, it is important to address these concerns and provide a more detailed explanation of how the proposed approach would be implemented and validated. This would require a more in-depth discussion of the technical aspects of NeSy reasoning and how it can be applied to real-world traffic scenarios.