scholarly-writing-refiner

name	scholarly-writing-refiner
description	Polishes academic English paragraph by paragraph, reviewing grammar, word choice, voice, coherence, and sentence structure. Outputs revision suggestions alongside polished text. Triggered by phrases like 'polish this paragraph,' 'check the grammar,' 'rewrite in academic English,' or keywords like manuscript editing, SCI polishing, and journal submission editing.
license	MIT

Scholarly Writing Refiner — Academic Paper English Polishing Knowledge Base

Helps users review and polish English papers paragraph by paragraph according to international academic journal standards. Covers grammar correction, academic word choice optimization, voice normalization, coherence strengthening, sentence variety, and more. Outputs revision suggestions along with the polished text.

Quick Start

Users only need to provide:

1. Text to polish: One or more paragraphs of English paper content
2. Paper type (optional): Journal article / Conference paper / Thesis or dissertation / Review article
3. Target journal/field (optional): e.g., Nature, IEEE, AAAI, Medicine, Computer Science, etc.
4. Polishing focus (optional): Comprehensive polish / Grammar only / Word choice only / Coherence only

Example:

"Polish this Introduction for me. The target venue is NeurIPS, and I'd like the language to sound more natural with better logical flow."

---

1. Review Dimensions Overview

Polishing is carried out across 5 dimensions. Each dimension is rated independently with specific revision suggestions:

Dimension	Label	Review Focus
Grammar	Grammar	Subject-verb agreement, tense, articles, prepositions, clause structure, punctuation
Word Choice	Word Choice	Academic register, precision, collocation, avoidance of colloquialisms
Voice	Voice & Tense	Active/passive voice selection, tense consistency
Coherence	Coherence & Cohesion	Intra-paragraph and inter-paragraph transitions, argumentation chain, use of signaling words
Sentence Structure	Sentence Structure	Sentence variety, balance of long and short sentences, coordination and subordination

---

2. Grammar Review Rules

2.1 Common Grammar Errors Checklist

Error Type	Incorrect Example	Correction	Explanation
Subject-verb disagreement	The results of the experiment shows...	The results of the experiment show...	Subject is "results" (plural)
Missing/misused article	We propose method to solve...	We propose a method to solve...	Singular countable nouns require an article
Dangling modifier	Using the proposed method, the accuracy was improved.	Using the proposed method, we improved the accuracy.	The implied subject of a participial phrase must match the main clause subject
Run-on sentence	The model performs well , it achieves 95% accuracy.	The model performs well**;** it achieves 95% accuracy. / The model performs well**. It** achieves 95% accuracy.	A comma cannot join two independent clauses
Incomplete comparison	Our method is more efficient.	Our method is more efficient than the baseline.	Comparatives require an explicit object of comparison
Broken parallelism	The system can detect, classify, and is able to segment...	The system can detect, classify, and segment...	Coordinated elements must share the same grammatical form
that/which confusion	The model which we proposed...	The model that we proposed...	Restrictive relative clauses use "that"
Irregular plurals	These phenomenon indicate...	These phenomena indicate...	Watch for irregular plural forms

2.2 Punctuation Rules

Rule	Correct Usage	Common Mistake
Serial comma (Oxford comma)	A, B**,** and C	A, B and C (the Oxford comma is recommended in academic writing)
Em dash	We used three models — A, B, and C — for comparison.	An em dash with spaces on both sides, or without spaces (depends on journal style)
Capitalization after colon	Capitalize if a complete sentence follows: The result is clear: The model outperforms...	Do not capitalize if a fragment follows
Quotation marks and periods	American style: period inside quotes. British style: period outside quotes.	Choose based on the target journal's regional convention
Abbreviation periods	e.g., i.e., et al., etc.	Note the comma: e.g., / i.e.,

2.3 Tense Guidelines by Paper Section

Section	Recommended Tense	Example
Abstract	Past tense (what was done) + present tense (conclusions)	"We proposed a method... The results show that..."
Introduction	Present tense (current knowledge/consensus) + past tense (prior work)	"Deep learning has become... Smith et al. demonstrated that..."
Methods	Past tense (experimental procedures)	"We trained the model on... The data were preprocessed..."
Results	Past tense (experimental findings)	"The model achieved 95% accuracy. Table 2 shows..."
Discussion	Present tense (interpreting significance) + past tense (citing results)	"This result suggests that... Our findings indicated that..."
Conclusion	Past tense (summarizing work) + present tense (contributions/significance)	"We proposed and evaluated... This work contributes to..."

---

3. Word Choice Optimization Rules

3.1 Colloquial → Academic Substitution Table

Colloquial	Academic Alternative	Context Notes
a lot of	numerous / a substantial number of / considerable	Choose based on what is being modified
get	obtain / acquire / achieve / attain	Choose based on collocation
show	demonstrate / illustrate / indicate / reveal	"demonstrate" emphasizes proof; "indicate" emphasizes suggestion
big / huge	substantial / significant / considerable
thing	factor / aspect / element / component
good	effective / favorable / advantageous / robust
bad	adverse / detrimental / suboptimal / inferior
use	employ / utilize / leverage / adopt	"utilize" is more formal than "use"; "leverage" emphasizes exploiting an advantage
about	approximately / roughly / circa	Use "approximately" for numerical descriptions
try	attempt / endeavor
look at	examine / investigate / analyze / explore
find out	determine / ascertain / identify / discover
go up / go down	increase / decrease / rise / decline
point out	highlight / emphasize / underscore
deal with	address / tackle / handle / mitigate
make sure	ensure / verify / confirm
kind of / sort of	somewhat / to some extent / partially
start / begin	initiate / commence / undertake
end / finish	conclude / terminate / complete
help	facilitate / enable / assist / contribute to
need	require / necessitate
can	is capable of / is able to / has the potential to	Avoid over-substitution — "can" is acceptable in academic writing

3.2 Vague → Precise Expression

Vague Expression	Precise Alternative	Notes
very good results	statistically significant improvement / a 12% increase in accuracy	Replace vague modifiers with concrete data
some researchers	Several studies (Chen et al., 2023; Li et al., 2024)	Replace vague references with specific citations
recently	In the past five years / Since 2020	Provide a time range
a few	three / a small number of (n=3)	Specify the quantity
it is known that	Prior work has established that (citation)	Support with a citation
this is important	This is critical for / This has significant implications for	Explain why it matters

3.3 Reducing Redundancy

Redundant Expression	Concise Version
in order to	to
due to the fact that	because / since
at the present time	currently / now
it is worth noting that	Notably, / Note that
it should be pointed out that	(state the content directly)
a total of 50 samples	50 samples
the vast majority of	most
in the event that	if
has the ability to	can
on a daily basis	daily
in close proximity to	near
take into consideration	consider
is in agreement with	agrees with
serves the function of	functions as

---

4. Voice Guidelines

4.1 Active vs. Passive Voice Selection

Scenario	Recommended Voice	Example
Describing the authors' actions	Active (We)	We trained the model using...
Describing general methods/established facts	Passive	The data were collected from...
Emphasizing the object of an action	Passive	The samples were analyzed using mass spectrometry.
Reporting results	Prefer active	Our method achieves 95% accuracy.
Describing equipment/materials	Passive	The solution was heated to 100°C.

4.2 Common Voice Issues

Issue	Incorrect Example	Correction
Overuse of passive	It was found by us that the results were improved by the method.	We found that our method improved the results.
Inconsistent person	The author proposes... We then evaluate...	Use "We" or "The authors" consistently
Meaningless passive	It can be seen that accuracy increases.	Accuracy increases. / The results show that accuracy increases.

4.3 Academic Person Conventions

Person	Use Case	Notes
We	Describing the authors' own work (most common)	Many journals accept "We" even for single-author papers
The authors	A more formal alternative	Some journals prefer this usage
I	Single-author theses and dissertations	Some journals do not accept this
One	Generic/hypothetical statements	Somewhat old-fashioned; less common in modern academic writing

---

5. Coherence and Cohesion Rules

5.1 Intra-Paragraph Signaling Words

Logical Relationship	Signal Words/Phrases	Example
Addition	Furthermore, Moreover, Additionally, In addition	Furthermore, our method generalizes well to unseen data.
Contrast	However, In contrast, Conversely, On the other hand, Nevertheless	However, this approach suffers from high computational cost.
Cause & Effect	Therefore, Consequently, As a result, Hence, Thus	Therefore, we adopt a two-stage training strategy.
Exemplification	For example, For instance, Specifically, In particular	Specifically, we focus on the image classification task.
Emphasis	Indeed, Notably, Importantly, It is worth noting that	Notably, the improvement is consistent across all datasets.
Concession	Although, Despite, Notwithstanding, While, Even though	Although the model is simple, it achieves competitive results.
Summary	In summary, To summarize, Overall, In conclusion	Overall, the proposed method outperforms existing baselines.
Qualification	Yet, Still, Nonetheless, That said	That said, there are several limitations to our approach.
Sequence	First, Second, Finally, Subsequently, Then	First, we preprocess the data. Subsequently, we train the model.
Condition	If, Provided that, Given that, Assuming that	Given that the dataset is imbalanced, we apply oversampling.

5.2 Inter-Paragraph Transition Patterns

Pattern	Description	Example Opening Sentence
Hook	The end of one paragraph leads into the next topic	"This raises the question of how to efficiently scale the model."
Recap	The next paragraph opens by revisiting the prior conclusion	"Having established the effectiveness of our approach, we now turn to..."
Contrast Bridge	Points out the shortcomings of the prior approach, introducing the current one	"While these methods achieve reasonable accuracy, they fail to address..."
Question Bridge	Uses a question to create a transition	"How can we overcome this limitation? In this section, we propose..."
Topic Sentence	The first sentence of each paragraph summarizes the core argument	"The key advantage of our method is its ability to..."

5.3 Common Coherence Problems

Problem	Description	Fix Strategy
Jumping argumentation	Leaping from A to C without the B step	Add intermediate reasoning steps or transitional sentences
Signal word overuse	Starting every sentence with However / Moreover	Reduce signal words; use sentence structure to convey logic
Signal word misuse	Using "Furthermore" to express contrast	Use "However" for contrast; "Furthermore" for addition
Overly long paragraphs	A single paragraph exceeding 8–10 sentences	Split into 2–3 paragraphs by argument point
Overly short paragraphs	A paragraph with only 1–2 sentences	Merge into a related paragraph or expand the discussion
Unclear reference	"This shows..." — what does "this" refer to?	"This result shows..." / "This finding indicates..."

---

6. Sentence Structure Optimization Rules

6.1 Strategies for Sentence Variety

Strategy	Original	Improved
Participial phrase opening	We use attention mechanism, and we improve accuracy.	Leveraging the attention mechanism, we improve accuracy.
Inversion for emphasis	The improvement is particularly notable in low-resource settings.	Particularly notable is the improvement in low-resource settings.
Appositive insertion	The model, which was proposed by Smith, achieves...	The model, proposed by Smith (2023), achieves...
Nominalization	We improved the model, and this led to...	The improvement of the model led to...
Parallel structure	The method is fast. It is also accurate. It is scalable too.	The method is fast, accurate, and scalable.
Fronted adverbial	Accuracy improved significantly when we added data augmentation.	With data augmentation, accuracy improved significantly.

6.2 Sentence Structure Problems to Avoid

Problem	Example	Fix
Overly long sentences (>40 words)	We trained the model on the dataset which was collected from ... and preprocessed using ... and then evaluated on ...	Split into 2–3 shorter sentences
Consecutive short sentences	The accuracy is high. The model is fast. It uses less memory.	Combine: The model achieves high accuracy with fast inference and low memory consumption.
Starting with There is/are	There are many studies that focus on...	Many studies focus on...
Overuse of It is...that cleft sentences	It is the attention mechanism that improves...	The attention mechanism improves...
Noun pile-ups	deep learning image classification model performance	the performance of a deep learning model for image classification

---

7. Section-Specific Polishing Guide

7.1 Abstract

• Length: 150–300 words (follow the target journal's requirements)
• Structure: Background (1–2 sentences) → Problem/Motivation (1 sentence) → Method (2–3 sentences) → Results (1–2 sentences) → Conclusion/Significance (1 sentence)
• Tense: Past tense for describing the work, present tense for conclusions
• Avoid: No citations, no abbreviations without first defining them in full, no figure or table numbers

7.2 Introduction

• Structure (classic "funnel" approach): Broad context → Specific problem → Existing methods and their limitations → Proposed method/contributions → Paper outline
• Key points: Each paragraph must have a clear topic sentence; be objective when reviewing prior work — do not disparage
• Common patterns:
  • "In recent years, ... has attracted increasing attention."
  • "Despite significant progress, ... remains a challenge."
  • "To address this issue, we propose..."
  • "The main contributions of this paper are as follows:"

7.3 Related Work

• Strategy: Organize by theme (not chronologically); within each group, arrange chronologically
• Key points: Explain how each work relates to the current paper; avoid mere listing — provide commentary
• Transitions: Connect each group with transitional sentences
• Common patterns:
  • "A closely related line of work focuses on..."
  • "In contrast to these approaches, our method..."
  • "Building upon the work of X, we extend..."

7.4 Methods

• Principle: Reproducibility — the reader should be able to replicate the experiment from the description alone
• Structure: Problem formulation → Overall framework → Detailed module descriptions → Training/optimization details
• Key points: Define all mathematical symbols upon first appearance; describe steps in execution order

7.5 Results / Experiments

• Structure: Experimental setup → Main results → Ablation studies → Analysis/Discussion
• Key points: Describe trends in text first, then reference tables/figures; avoid repeating numbers already shown in tables or figures
• Common patterns:
  • "As shown in Table X, our method outperforms..."
  • "We observe a consistent improvement of X% across..."
  • "The ablation study reveals that..."

7.6 Discussion

• Content: Interpret the significance of results → Compare with prior work → Limitations → Future directions
• Key points: Do not shy away from limitations; the discussion should go beyond the results themselves and explore broader implications

7.7 Conclusion

• Length: Typically one paragraph, 150–250 words
• Structure: Summarize the method → Core findings → Significance/Contributions → Future work
• Avoid: Do not introduce new information or data; do not simply repeat the Abstract

---

8. Polishing Output Format

For each paragraph of text provided by the user, output in the following format:

### Original
[User's original text]

### Review

| Dimension | Rating | Key Issues |
|-----------|--------|-----------|
| Grammar | ✓ Good / △ Needs improvement / ✗ Significant issues | Brief description |
| Word Choice | ✓ / △ / ✗ | Brief description |
| Voice | ✓ / △ / ✗ | Brief description |
| Coherence | ✓ / △ / ✗ | Brief description |
| Sentence Structure | ✓ / △ / ✗ | Brief description |

### Detailed Changes
1. **Original**: "..."
   **Revised**: "..."
   **Reason**: [Specific rationale citing the rules above]

2. ...

### Polished Version
[Complete polished paragraph]

---

9. Domain-Specific Notes

Different disciplines have their own writing conventions. Respect field-specific norms when polishing:

Field	Characteristics	Notes
Computer Science	Active voice ("We") is common	More colloquial phrasing is tolerated (e.g., "we run"); algorithm descriptions must be precise
Medicine/Biology	Passive voice predominates	"Patients were randomized..."; terminology must conform to MeSH standards
Physics	Concise, equation-driven	Mathematical derivations must be rigorous; "one can show that..." is common
Social Sciences	Frequent use of hedging	"may," "might," "suggests"; avoid overly absolute statements
Engineering	Results-oriented	Emphasis on performance metrics and experimental validation

---

10. Agent Behavior Guide

When the user submits text for polishing, follow this workflow:

1. Confirm details: Paper type, target journal/conference (if any), polishing focus
2. Identify the section: Determine which part of the paper the text belongs to and apply the corresponding section guidelines
3. Five-dimension review: Check systematically: Grammar → Word Choice → Voice → Coherence → Sentence Structure
4. Annotate sentence by sentence: Provide a reason for every change, citing the specific rule
5. Output the polished version: Deliver the complete polished text
6. Summarize recommendations: Highlight the main categories of issues and directions for improvement

Core Principles:

• Preserve the author's original meaning and argumentation logic; do not alter technical content
• Minimize changes — if a single word can be changed instead of a whole sentence, change only the word; if a sentence can be changed instead of a whole paragraph, change only the sentence
• When uncertain whether something is an error, present it as a suggestion rather than making the change outright
• Defer to the author's terminology unless it is clearly incorrect
• Do not modify content the user has marked as "please keep"