Why: Corrects LaTeX syntax (options go before class name), fixes broken email hyperlink (empty mailto), and uses proper \today macro.
Main Body Text Major Rewrite
Original
Thank you once again for the constructive feedback provided during this second review process. We have carefully revised the manuscript, addressing each of the reviewers' comments and have made the corresponding minor improvements clarity and quality of writing of the work. The manuscript now includes precise explanations on how AI futher enhance systematic reviews and its limitations, and better theoretical anchoring.
A complete, point-by-point response is provided in the "Response to Reviewers" document, and all modifications have been highlighted in the revised manuscript as requested.
We respectfully submit the revised version for your consideration and hope that it will now meet the standards for publication in Financial Innovation.
Improved
Thank you for the constructive feedback provided during this second review. We have carefully addressed each comment through targeted revisions focusing on consistency, theoretical anchoring, and the strengthening of concept-centric synthesis. In particular, the manuscript now includes a structured synthesis using the Theory–Context–Method framework, improved alignment between research questions and screening methodology, and explicit bounding of the study's scope as a framework contribution demonstrated through a financial narratives case study.
A complete point-by-point response is provided in the accompanying document, and all modifications have been highlighted in the revised manuscript.
We respectfully submit the revised version for your consideration.
Why (5 changes):
1. "minor improvements clarity and quality" → "targeted revisions" — Fixes grammatical error (missing "to the") AND replaces dishonest "minor" framing with honest characterization
2. "AI futher enhance" → eliminated — Fixes typo and removes vague claim
3. Adds specific content: TCM framework, RQ alignment, framework scoping — Shows the editor what was actually done
4. "as requested" removed — Unnecessary; slightly passive-aggressive tone
5. "hope that it will now meet the standards" removed — Reads as needy; confidence is conveyed by brevity
Response Letter: General Statement
Complete Rewrite Major Rewrite
Original (~45 words)
We thank the Editor and the reviewers for their detailed and constructive feedback. We acknowledge that this revision focuses on resolving the remaining issues related to consistency, positioning, and the strengthening of the systematic literature review contribution.
All remaining points have been carefully addressed and are detailed below.
Improved (~75 words)
We thank the Editor and the reviewers for their continued engagement with this work. The paper proposes a reusable algorithmic framework for systematic literature reviews and demonstrates it through a case study on financial narratives.
This revision addresses three categories of feedback: (1) consistency corrections between research questions, screening criteria, and manuscript text; (2) the addition of a concept-centric synthesis using the Theory–Context–Method framework; and (3) explicit scoping of the framework's current implementation choices, in particular the use of Scopus as the bibliographic source and the role of researcher-defined screening statements.
All points are addressed below.
Why (the single most important strategic change):
1. Establishes "framework + case study" framing in the very first sentence the editor reads. This colors how Comments 3–4 are interpreted: the Scopus and subjectivity limitations are case-study implementation choices, not framework flaws.
2. Enumerates three specific revision categories instead of vague "remaining issues." Category (3) pre-frames Comments 3–4 as "implementation choices" before the editor encounters them.
3. Removes "All remaining points have been carefully addressed" — functionally the same overclaim as the Conclusion. Replaced with neutral "All points are addressed below."
Reviewer 2 — Comment 1: Grossman–Stiglitz (1980)
Minor Polish Preserved
Original (~55 words)
We thank the reviewer for pointing out this inconsistency. We have corrected the reference throughout the manuscript by replacing the SSRN entry, which only cited Grossman (1980), with the original published article by Grossman and Stiglitz (1980) in The American Economic Review. The in-text citation and the corresponding bibliography entry have been updated accordingly.
Improved (~50 words)
We thank the reviewer for identifying this oversight. The reference has now been corrected throughout: the SSRN entry citing only Grossman (1980) has been replaced with the original published article by Grossman and Stiglitz (1980) in The American Economic Review. Both the in-text citation and the bibliography entry have been updated accordingly.
Why: "pointing out this inconsistency" softened to "identifying this oversight" — acknowledges it was an oversight (honest) without drawing more attention to it. Otherwise substantively identical.
Reviewer 2 — Comment 2: RQ2 Consistency
Minor Polish Preserved
Original (~175 words)
We thank the reviewer for this careful reading.
To ensure full consistency with the revised formulation of RQ2, we reviewed the manuscript and corrected all remaining references that reflected the previous focus on financial market dynamics.
In particular, screening statements that previously referred to "understanding" or "enhancing" market dynamics have been reformulated to align with the current objective of the review, namely the conceptualization and modeling of financial narratives.
In addition, we clarified the role of search terms related to market dynamics in the screening procedure. [...]
Finally, the keywords and remaining minor inconsistencies have been updated [...]
Improved (~165 words)
Substantively identical. Minor compression only. Same structure: (1) reviewed and corrected, (2) screening statements reformulated, (3) search terms clarified, (4) keywords updated.
Why: This response was already strong — thorough, specific, enumerates concrete changes. Preserved with only minor wording tightening.
Reviewer 2 — Comment 3: Single-Source (Scopus)
Dramatic Cut Major Cut −73%
Original (~280 words, 4 paragraphs)
We thank the reviewer for raising this point. We agree that reliance on a single bibliographic source is an important limitation of the current implementation. As already stated in the manuscript, Scopus was retained because it offers a practical compromise between accessibility, metadata consistency, and scientific quality control. In the context of an algorithmic screening pipeline, this balance is important: abstracts, keywords, document types, and indexing fields are relatively standardized [...]
We would also like to clarify that this choice should not be interpreted as claiming that Scopus is exhaustive, nor that it is intrinsically superior to all other sources [...]
As also discussed in the revised manuscript, a natural extension of the framework would be to rely on OpenAlex [...] Beyond this database-level extension, an even more ambitious improvement would be to move away from traditional keyword-constrained search endpoints and adopt vector-based retrieval directly at the corpus level [...]
For the present paper, however, we chose to keep the Scopus-based implementation because it provides a controlled and reproducible baseline for demonstrating the selection framework.
Improved (~75 words, 1 paragraph)
We agree that single-source reliance is a limitation of the current implementation, and this is acknowledged in the manuscript. Scopus was chosen for this case study because its standardized metadata enables reproducible algorithmic screening. The framework itself is source-agnostic; extending it to additional databases or unified APIs such as OpenAlex is a direction we discuss in the paper. For the present work, Scopus provides a controlled baseline for demonstrating the selection pipeline.
Why (the highest-impact framing change):
1. At R2, the reviewer already heard the full argument at R1. Repeating it at greater length signals anxiety. Brevity signals confidence.
2. Original listed 2 separate future-work items (OpenAlex AND vector-based retrieval). Multiple "future work" promises suggest premature publication. Reduced to 1 bounded mention.
3. "The framework itself is source-agnostic" is the key strategic phrase — frames Scopus as a case-study choice, not a framework limitation.
4. Removed the defensive "should not be interpreted as claiming" paragraph entirely — explaining what you're NOT claiming reads as protesting too much.
5. 4 paragraphs of justification for NOT doing something → 1 paragraph that acknowledges, bounds, and moves on.
Reviewer 2 — Comment 4: Subjectivity in Inclusion Criteria
Dramatic Cut Major Cut −78%
Original (~320 words, 3.5 paragraphs)
We thank the reviewer for raising this point. We agree that the use of researcher-defined statements introduces a degree of subjectivity [...] several design choices were implemented to mitigate this subjectivity and reduce dependence on arbitrary wording. First, the embedding model used (text-embedding-3-small) is a large-scale, general-purpose model [...]
Second, we introduced an explicit rephrasing step using a large language model. Each statement is automatically reformulated into multiple semantically equivalent variants [...]
Third, the six statements used in the screening process were intentionally designed to be broad and complementary [...]
More broadly, this limitation is partly inherited from the constraints of current literature databases [...] a more robust setup would consist of indexing the entire corpus as a vector database [...]
Finally, we agree that a formal validation protocol would further strengthen the study [...] implementing such a validation framework would require a substantial annotation effort and would constitute a research contribution on its own [...]
Improved (~70 words, 1 paragraph)
We acknowledge that researcher-defined screening statements introduce subjectivity, as is inherent in any semantic retrieval approach without a predefined benchmark. The framework mitigates this through the use of a general-purpose embedding model, automatic LLM-based rephrasing of statements into multiple semantic variants, and the use of broad, complementary statements that function as structured semantic filters rather than rigid rules. These design choices and the limitation itself are discussed in the manuscript.
Why:
1. Same logic as Comment 3: brevity = confidence at R2.
2. The 3 mitigations are compressed into a single clause instead of 3 separate paragraphs. The reviewer already read the paper.
3. Critical cut:"would constitute a research contribution on its own" is removed. This sentence told the editor: "we know validation is needed and we didn't do it." Better to simply bound scope and move on.
4. Vector database future work removed (was the 4th "future work" item across Comments 3–4).
5. "These design choices and the limitation itself are discussed in the manuscript" — directs the editor to the paper instead of re-teaching the methods in the response letter.
[...] This is a valid concern, particularly in research where the boundary between relevance levels is not clearly defined.
In our use case, this design choice was not applied blindly. [...] manual inspection of papers [...] confirmed that a large share of these papers did not directly address the research questions [...]
More importantly, this aspect should be understood as a configurable component [...] (i) binary classification [...] (ii) three-tier classification [...] The appropriate choice depends on several factors, including the breadth of the research field, the maturity and density of the literature, and the specific objectives of the review. In well-established and highly populated domains, introducing an intermediate class [...] Conversely, in emerging or sparsely populated fields, retaining these papers may be preferable [...]
Improved (~175 words)
Same core argument preserved: manual inspection confirmed boundary papers were peripheral; framework offers configurable binary or three-tier classification.
Trimmed: the field-maturity discussion (well-established vs. emerging domains) shortened to one sentence: "The appropriate choice depends on the breadth and density of the literature under review."
Why: Already well-handled. Minor tightening only. The field-maturity argument was thoughtful but added length without changing the core point.
Reviewer 5 — Major: TCM Synthesis
Unchanged Preserved
Original (~195 words)
Textbook responsive revision: new subsection added using TCM framework (Paul & Criado 2020), concept matrix constructed across three dimensions, research gaps derived from matrix absences.
Improved (~190 words)
Substantively identical. This was the strongest response in the original letter. Preserved in full.
Why: Excellent as-is. The reviewer asked for TCM and a concept matrix; the authors delivered. No changes needed.
Reviewer 5 — Minor 1: AI in Synthesis
Slight Trim Minor
Original (~175 words)
[... same first 3 paragraphs ...]
That said, the results suggest that combining vector-based retrieval with adapted reranking techniques provides a scalable foundation that could, in future work, be extended toward more advanced synthesis pipelines, for instance within vector database environments where similarity scoring and clustering can be applied at scale.
Improved (~145 words)
Same first 3 paragraphs. Drops the final paragraph about vector-based synthesis extension.
Why: The dropped paragraph discussed future synthesis extensibility — it was forward-looking but added yet another "future work" item. The response is already honest and well-scoped without it.
Identical content. Year filter, screening statement 6, Section 2.1 clarification. Shows unsolicited diligence — important counter-evidence to the carelessness pattern.
Conclusion
Rewritten Major Rewrite
Original (~35 words)
We believe that this revision addresses all remaining concerns raised by the reviewers. The manuscript has been corrected for consistency issues and strengthened in its synthesis and gap-identification components.
We thank the reviewers again for their valuable feedback.
Improved (~45 words)
This revision addresses the reviewers' comments through targeted corrections, a strengthened synthesis component, and explicit scoping of the framework's current implementation boundaries. The manuscript has been corrected for consistency, enriched with a concept-centric synthesis using the TCM framework, and revised to clearly delineate the scope of the present case-study implementation.
We thank the reviewers for their valuable feedback throughout the review process.
Why:
1. "addresses all remaining concerns" is factually false — Comments 3 and 4 are acknowledged limitations, not resolved concerns. The editor knows this. Claiming otherwise invites a credibility check the authors cannot pass.
2. "We believe that" hedges a false claim — makes it both wrong and weak. Removed.
3. Replacement names what was actually done (corrections + synthesis + scoping) without claiming completeness on unresolved items.
4. "implementation boundaries" subtly references the Scopus and subjectivity limitations without re-opening them.
Summary Table
Section
Original
Improved
Change
Type
Cover letter body
~75 words
~80 words
Rewritten: grammar fixed, honest framing
Major
General Statement
~45 words
~75 words
Framework+case-study framing established
Major
Comment 1 (Grossman–Stiglitz)
~55 words
~50 words
Minor wording polish
Preserved
Comment 2 (RQ2)
~175 words
~165 words
Minor compression
Preserved
Comment 3 (Scopus)
~280 words
~75 words
−73% — 4 paragraphs → 1
Major
Comment 4 (Subjectivity)
~320 words
~70 words
−78% — 3.5 paragraphs → 1
Major
Comment 5 (Medium-relevance)
~195 words
~175 words
Minor trim
Minor
Rev 5 Major (TCM)
~195 words
~190 words
Preserved
Preserved
Rev 5 Minor 1 (AI synthesis)
~175 words
~145 words
Dropped final future-work paragraph
Minor
Rev 5 Minor 2 (Typos)
~55 words
~55 words
Preserved
Preserved
Additional Corrections
~95 words
~95 words
Preserved
Preserved
Conclusion
~35 words
~45 words
Overclaim removed, honest framing
Major
Total response letter
~1,625 words (5 pages)
~1,130 words (4 pages)
−30% overall
Future-Work Items Comparison
Original (4 items)
Improved (1 item)
1. OpenAlex multi-source extension
2. Vector-based retrieval at corpus level
3. Formal validation protocol for screening statements
4. AI-assisted synthesis pipeline
1. OpenAlex extension (mentioned as discussed in paper)
Why: 4 "future work" items suggests the paper is a partial implementation published prematurely. 1 bounded mention suggests a complete contribution with a natural extension path.