← Back to The Index Final Review Submission #2
Final Review

Review Nº 02

Scoping Review
AuthorsMatteo Pallomo, Eva Ledovskaia, Andrea Barbantani, Alessio Quaranta, Marco Farano, Elena Ruberto, Fiamma Pia Paternò, Michele Barale, Federico Ciociola and Flora D'Angelo
27/30
Score
Building on an already strong original (24/30), the revision delivers meaningful improvements — a transparent PRISMA-backed methodology, resolution of the Weighted Performance ambiguity into a coherent five-dimension model, integrated privacy/governance, added Abstract and Conclusion, and a complete factor-to-source mapping — but loses points because the ADAS use case remains descriptive rather than evaluative, the gaps section still conflates self-critique with literature gaps, and no operational trust-recovery protocol is provided.

The Pros

7 Items
+
Clear and traceable methodology: Scopus query, timeframe, 599→265→47 funnel, PRISMA diagram, and explicit inclusion/exclusion criteria now documented.
+
The ambiguous "Weighted Performance" construct was eliminated and replaced by a clean dynamic-alignment model with five interdependent macro-categories.
+
Abstract and Conclusion are now present and informative, fixing a critical structural gap.
+
Privacy, governance, and ethical safeguards are integrated into the Objective System Properties layer rather than left as appendix material.
+
Comprehensive factor-to-reference mapping (Table 3) gives every coded concept clear source attribution, addressing replicability concerns.
+
Figure 1 has been redesigned and is no longer marred by typographical errors; the framework reads professionally.
+
Adaptive Trust Dynamics now explicitly models asymmetry, feedback loops, trust transfer, and mediation effects, strengthening the temporal account.

The Cons

7 Items
The ADAS use case continues to rephrase the framework with domain vocabulary instead of evaluating a real or simulated system on Measured / Perceived / Contextual axes.
§12 "Gaps and Future Work" still leans toward authors' methodological self-critique (English-only, citation bias, engineering background) rather than gaps in the literature itself (longitudinal trust recovery, high-stakes domains, cross-cultural studies).
No concrete trust-repair / recovery protocol is provided despite this being a specific consensus suggestion; trust repair is named but not operationalized.
Ethical breadth (dignity, autonomy, professional responsibility) and social-science depth (cognitive biases, trust heuristics) remain thin relative to the technical-governance treatment.
A minor governance/accountability typo remains ("trustoor and trustee" in Table 3 entries for [38] and [4]).
The use of [13] (Kim et al. 2022) to support "continuous learning / system evolution" is a stretch — that source is about service robots and social exchange theory, not online learning — and may signal residual mis-mapping in the coding scheme.

Suggested Changes

12 Pointers
01
High
Location
§1–§12 (overall length)
Issue
Main body still spans 5 pages, violating the 4-page double-column limit
Suggested Fix
Requirement relaxed for the final submission
02
High
Location
§10 (ADAS use case)
Issue
Use case rephrases the framework rather than evaluating a system
Suggested Fix
Add a small evaluation block: pick one concrete AI requirement-generation tool (real or simulated), score it qualitatively across Measured / Perceived / Contextual / Adaptive dimensions, and report observed (or hypothesized) miscalibration outcomes
03
High
Location
§12 (Gaps and Future Work)
Issue
Mixes authors' selection biases with genuine literature gaps; literature gaps are underdeveloped
Suggested Fix
Split into two subsections: "Limitations of this review" (English-only, citation cutoff, engineering bias) and "Open research gaps in the literature" (longitudinal trust-recovery studies, high-stakes domain coverage, cross-cultural validation, scarcity of practical sources)
04
High
Location
§9 (Adaptive Trust Dynamics)
Issue
"Trust repair" is named but not operationalized; reviewers explicitly asked for a recovery protocol
Suggested Fix
Add a short paragraph or bullet list of concrete repair actions (e.g., post-failure explanation, staged re-introduction with reduced autonomy, transparent error logs, calibration retraining) and cite them
05
Medium
Location
§5 / §8 (ethical and social depth)
Issue
Ethics still framed mainly as governance/privacy; dignity, autonomy, professional identity not developed
Suggested Fix
Promote "professional identity threat" and "accountability" out of the appendix table into the main text of §8, and add a one-paragraph treatment of cognitive biases / trust heuristics (anchoring, automation bias) in §6
06
Medium
Location
§3 (Methods)
Issue
The 600-paper citation-threshold cutoff is justified pragmatically but acknowledged as a bias in §12 only
Suggested Fix
Add one sentence in §3 noting that this citation-based ranking systematically under-represents 2024–2026 work, so readers see the limitation where the choice is made
07
Medium
Location
§3 (Methods)
Issue
Single-reviewer abstract screening with a "locking mechanism" — no inter-rater agreement reported
Suggested Fix
Either (a) report a small double-screening sample with Cohen's κ, or (b) explicitly state this as a methodological limitation and reference PRISMA-ScR conventions for transparency
08
Medium
Location
§4 + Figure 1
Issue
Caption says "Four main dimensions" but the framework defines five (Adaptive Trust Dynamics is the fifth)
Suggested Fix
Fix the caption to read "Five interdependent dimensions…" and ensure the figure visually distinguishes the four static layers from Adaptive Trust Dynamics as the over-time process
09
Low
Location
Appendix B, Table 3
Issue
"trustoor and trustee" typo in entries for [38] and [4]; "Objective Metrics" used inconsistently with "Objective Sys. Properties" elsewhere
Suggested Fix
Correct to "trustor and trustee"; unify the layer label to "Objective System Properties" throughout the table
10
Medium
Location
§9 + Reference [13]
Issue
[13] (Kim et al., service robots / social exchange theory) is cited to support "continuous learning / system evolution" — weak fit
Suggested Fix
Replace with a source that actually treats online learning or model evolution, or recast the claim to match what [13] supports (relational/social-exchange evolution of trust)
11
Low
Location
§6 (Subjective Perception)
Issue
"Perceived performance may diverge from actual performance" is the cleanest replacement for the old Weighted Performance idea but is mentioned only briefly
Suggested Fix
Expand by one or two sentences with a worked mini-example (e.g., a high-accuracy classifier perceived as untrustworthy because of opaque outputs) to make the calibration-gap concept concrete
12
Low
Location
§10.1 (ADAS contextual factors)
Issue
"Colleagues might judge him as lazy and/or incompetent" is informal phrasing in an otherwise academic section
Suggested Fix
Rewrite in academic register, e.g., "Peer perception and concerns about being seen as overly reliant on automation introduce a social dimension to trust formation"
Back to The Index
Score · 27/30
Good · Not · Done
Pros / Cons / Pointers
Final Review · Submission #2 The Index Grandi Sfide · 2026