psyclaw 0.27.23 → 0.28.1

package/vendor/ars/academic-paper/references/abstract_writing_guide.md

@@ -0,0 +1,169 @@
+# Abstract Writing Guide
+
+Used by `abstract_bilingual_agent`.
+
+## Abstract Types
+
+### Structured Abstract
+Contains explicit labeled sections. Required by many journals in social sciences and medicine.
+
+**Sections**: Background, Purpose/Objective, Method, Results/Findings, Conclusion/Implications
+
+### Unstructured Abstract
+A single flowing paragraph without labels. Common in humanities and some social sciences.
+
+**Flow**: Context → Problem → Purpose → Method → Key Findings → Implications
+
+### Extended Abstract
+Longer (500-1,000 words), used for conference submissions. May include brief literature review and preliminary results.
+
+## English Abstract Guidelines
+
+### Word Count
+- Standard: 150-250 words (check journal requirements)
+- Conference: 200-500 words (check CFP)
+- Dissertation: up to 350 words
+
+### Structure (5-Component Model)
+
+#### Component 1: Background (1-2 sentences)
+Establish context and identify the problem.
+
+**Patterns**:
+- "[Topic] has become increasingly important because..."
+- "Despite growing interest in [topic], little is known about..."
+- "Recent developments in [field] have raised questions about..."
+
+**Avoid**:
+- Starting with "This paper..." (too abrupt)
+- Generic statements ("Education is important")
+- Overly long historical context
+
+#### Component 2: Purpose (1 sentence)
+State the specific objective or research question.
+
+**Patterns**:
+- "This study examines [what] in [context]."
+- "The purpose of this research is to [verb] [object]."
+- "This paper proposes [framework/model] for [application]."
+
+#### Component 3: Method (1-2 sentences)
+Describe the approach, data, and analysis.
+
+**Patterns**:
+- "Using [method], this study analyzed [data] from [source]."
+- "A [design] approach was employed, involving [participants/data]."
+- "Data were collected through [instrument] and analyzed using [technique]."
+
+#### Component 4: Findings (2-3 sentences)
+Present the key results — be specific.
+
+**Patterns**:
+- "The results indicate that [finding 1]. Additionally, [finding 2]."
+- "Three key findings emerged: (a) [finding 1], (b) [finding 2], and (c) [finding 3]."
+- "The analysis revealed [main finding], with [specific metric/detail]."
+
+**Include**:
+- Specific numbers when available (percentages, effect sizes)
+- The most important findings (not all findings)
+
+**Avoid**:
+- "Results will be discussed" (the abstract IS the discussion)
+- Vague findings ("significant results were found")
+
+#### Component 5: Implications (1-2 sentences)
+State the significance, practical implications, or recommendations.
+
+**Patterns**:
+- "These findings have implications for [practice/policy/theory]."
+- "The results suggest that [stakeholders] should [action]."
+- "This research contributes to [field] by [contribution]."
+
+### Example (English, Education)
+
+> Declining enrollment poses existential challenges for private higher education institutions in Taiwan, yet institutional responses remain poorly understood. This study examines the strategic adaptation patterns of 12 private universities experiencing enrollment declines exceeding 20% between 2018 and 2023. Using a multiple case study design, we analyzed institutional documents, enrollment data, and 36 semi-structured interviews with administrators. Three distinct adaptation strategies emerged: program consolidation (n = 5), niche specialization (n = 4), and merger pursuit (n = 3). Institutions adopting niche specialization demonstrated the highest enrollment recovery rates (mean = 12.3%). However, successful adaptation was contingent upon early action — institutions that initiated strategic changes within two years of enrollment decline showed significantly better outcomes than late movers (p < .01). These findings suggest that private institutions should adopt proactive monitoring systems and consider niche specialization as a primary survival strategy.
+>
+> **Keywords**: higher education, enrollment decline, private universities, institutional strategy, Taiwan
+
+## Traditional Chinese Abstract Guidelines
+
+### Word Count
+- Standard: 300-500 characters
+- Conference: 300-800 characters
+- Dissertation: 500-1,000 characters
+
+### Structure (5-Component Model)
+
+#### Component 1: Research Background (1-2 sentences)
+**Patterns**:
+- "As... has become an important issue for..."
+- "In the context of..., ... faces the challenge of..."
+- "Although... has received widespread attention, ... still lacks systematic research."
+
+#### Component 2: Research Purpose (1 sentence)
+**Patterns**:
+- "This study aims to explore..."
+- "This paper takes... as subjects and analyzes the impact of..."
+- "The purpose of this study is to construct a framework for..."
+
+#### Component 3: Research Method (1-2 sentences)
+**Patterns**:
+- "This study employs... research method, with... as subjects, collecting data through..."
+- "Using... to analyze... data."
+
+#### Component 4: Research Findings (2-3 sentences)
+**Patterns**:
+- "The results found: (1)...; (2)...; (3)..."
+- "The analysis results show that... Furthermore,..."
+
+#### Component 5: Research Significance (1-2 sentences)
+**Patterns**:
+- "This study has both theoretical and practical contributions to..."
+- "The results can serve as a reference for... and provide specific recommendations for..."
+
+### Example (Chinese, Higher Education Field)
+
+> Amid the trend of declining birth rates, Taiwan's private universities face enrollment difficulties, yet existing research on institutional response strategies remains limited. This study aims to explore the strategic adaptation patterns of 12 private universities that experienced enrollment declines exceeding 20% between 2018 and 2023. The study employs a multiple case study design, analyzing institutional development plans, enrollment data, and conducting 36 semi-structured interviews with administrators. Three primary adaptation strategies were identified: program consolidation (5 institutions), niche specialization (4 institutions), and merger pursuit (3 institutions). Among these, institutions adopting niche specialization demonstrated the highest enrollment recovery rates (mean = 12.3%). The study also found that timing of adaptation is critical to success -- institutions that initiated strategic adjustments within two years of enrollment decline showed significantly better outcomes than late movers. The results suggest that private universities should establish proactive monitoring mechanisms and adopt niche specialization as a primary sustainability strategy.
+>
+> **Keywords**: higher education, enrollment decline, private university, institutional strategy, Taiwan
+
+## Keywords Selection
+
+### English Keywords
+1. **Core concepts** — main variables or constructs (2-3)
+2. **Context** — geographical, institutional, or temporal (1-2)
+3. **Method** — if distinctive (0-1)
+4. **Field** — discipline or sub-field (1)
+
+**Rules**:
+- Lowercase (unless proper nouns)
+- Complement the title (don't repeat title words verbatim)
+- Use established terms (check journal's keyword list if available)
+- 5-7 keywords total
+
+### Chinese Keywords
+1. **Core concepts** — main variables or constructs (2-3)
+2. **Research context** — geographical, institutional, or temporal (1-2)
+3. **Research method** — if distinctive (0-1)
+4. **Academic field** — discipline or sub-field (1)
+
+**Rules**:
+- Use formal academic terminology
+- Avoid completely duplicating the title
+- May reference the National Central Library Chinese Subject Headings
+- 5-7 keywords
+
+## Bilingual Abstract Quality Checklist
+
+| Check | ✓ |
+|-------|---|
+| Both abstracts cover all 5 components | |
+| English: 150-300 words | |
+| Chinese: 300-500 characters | |
+| Abstracts are independently written (not translated) | |
+| Key findings match between languages | |
+| Quantitative data consistent between versions | |
+| Keywords: 5-7 per language | |
+| Keywords complement (not duplicate) the title | |
+| No citations in the abstract | |
+| No abbreviations undefined in the abstract | |

package/vendor/ars/academic-paper/references/academic_writing_style.md

@@ -0,0 +1,188 @@
+# Academic Writing Style Guide
+
+Used by `draft_writer_agent` and `peer_reviewer_agent`.
+
+## Core Principles
+
+### 1. Precision
+- Use the most specific term available
+- Define technical terms on first use
+- Avoid ambiguous pronouns ("this," "it") without clear antecedents
+
+### 2. Conciseness
+- Eliminate filler words and redundant phrases
+- One idea per sentence (or clearly connected ideas)
+- Prefer short sentences for complex ideas
+
+### 3. Objectivity
+- Base claims on evidence, not opinion
+- Use hedging for uncertain claims
+- Acknowledge limitations and alternative interpretations
+
+### 4. Formality
+- Use full forms ("do not" over "don't")
+- Use formal academic vocabulary; reserve colloquialisms and slang for informal writing
+- Use third person unless discipline conventions allow first person
+
+## Register Adjustment by Discipline
+
+### Sciences (Natural, Applied)
+```
+Register: Formal, impersonal, method-focused
+Voice: Passive voice common ("was measured," "were analyzed")
+Terminology: Precise measurements, SI units, statistical notation
+Example: "The sample was heated to 350°C for 2 hours, yielding a conversion rate of 87.3% (SD = 2.1)."
+```
+
+### Social Sciences
+```
+Register: Formal, theory-informed, participant-aware
+Voice: Active voice encouraged, first person for researcher decisions
+Terminology: Theoretical constructs, operationalized variables
+Example: "We employed semi-structured interviews to explore how participants understood institutional change (N = 24)."
+```
+
+### Humanities
+```
+Register: Formal, argument-driven, interpretive
+Voice: First person acceptable for arguments, active voice
+Terminology: Close reading vocabulary, theoretical language
+Example: "I argue that the text's spatial metaphors reveal an underlying anxiety about institutional permanence."
+```
+
+### Engineering / CS
+```
+Register: Formal, problem-solution oriented, specification-precise
+Voice: Passive common for methods, active for contributions
+Terminology: Technical specifications, performance metrics
+Example: "The proposed algorithm achieves O(n log n) complexity, outperforming the baseline by 34% on the benchmark dataset."
+```
+
+### Education
+```
+Register: Formal, practice-oriented, stakeholder-aware
+Voice: Active voice, first person for reflexive practice
+Terminology: Pedagogical concepts, assessment language
+Example: "The intervention improved student metacognitive awareness, as evidenced by a significant increase in self-regulation scores (t(45) = 3.21, p = .002, d = 0.72)."
+```
+
+### Medicine / Health
+```
+Register: Formal, evidence-hierarchy conscious, clinical precision
+Voice: Passive for methods, active for findings
+Terminology: Clinical terms, diagnostic criteria, statistical reporting
+Example: "Patients receiving the intervention showed a 40% reduction in readmission rates (RR = 0.60, 95% CI [0.45, 0.80], p = .001)."
+```
+
+## Hedging and Strength Language
+
+### Hedging (for uncertain or qualified claims)
+| Strength | Hedging Devices | Example |
+|----------|----------------|---------|
+| Weak | may, might, could, possibly | "This may suggest a correlation." |
+| Moderate | suggests, indicates, appears | "The data suggest a positive trend." |
+| Strong | demonstrates, establishes, confirms | "The evidence demonstrates a clear link." |
+
+### When to Hedge
+- Results that need replication
+- Causal claims from correlational data
+- Generalizations from limited samples
+- Interpretations with alternative explanations
+
+### When NOT to Hedge
+- Reporting factual data: "The response rate was 78%." (not "appeared to be")
+- Describing methodology: "We used thematic analysis." (not "we attempted to use")
+- Well-established facts: "Earth orbits the Sun." (not "may orbit")
+
+## Transition Words and Phrases
+
+### Addition
+moreover, furthermore, in addition, additionally, similarly, likewise
+
+### Contrast
+however, nevertheless, in contrast, on the other hand, conversely, whereas
+
+### Cause/Effect
+therefore, consequently, as a result, thus, hence, accordingly
+
+### Example
+for example, for instance, specifically, in particular, such as, namely
+
+### Sequence
+first, second, third, subsequently, finally, meanwhile
+
+### Summary
+in summary, to conclude, overall, taken together, in short
+
+### Concession
+although, despite, while, granted that, notwithstanding
+
+## Paragraph Construction
+
+### Standard Academic Paragraph (TEEL)
+1. **T**opic sentence — states the paragraph's main point
+2. **E**vidence — data, citations, examples that support the point
+3. **E**xplanation — interpret the evidence, connect to argument
+4. **L**ink — connect to the next paragraph or back to thesis
+
+### Example
+> **[T]** AI-assisted quality assurance has shown promise in improving evaluation consistency across institutions. **[E]** Smith (2024) found that institutions using AI tools reported a 23% reduction in inter-rater variance, while Chen and Wang (2023) documented improved agreement on scoring rubrics (κ = 0.82 vs. 0.64). **[E]** These findings suggest that algorithmic assistance can mitigate the subjective biases inherent in human evaluation, particularly when assessors have varying levels of experience. **[L]** However, the reliance on AI tools also raises concerns about the loss of contextual judgment, which the following section addresses.
+
+## Common Style Errors
+
+### Wordiness
+| Wordy | Concise |
+|-------|---------|
+| in order to | to |
+| due to the fact that | because |
+| a large number of | many |
+| at the present time | currently / now |
+| it is important to note that | notably |
+| in the event that | if |
+| has the ability to | can |
+| with regard to | regarding / about |
+| in spite of the fact that | despite / although |
+| conduct an investigation of | investigate |
+
+### Vague Language
+| Vague | Precise |
+|-------|---------|
+| "many studies" | "several studies (e.g., Chen, 2023; Smith, 2024)" |
+| "a significant impact" | "a 23% increase in retention rates" |
+| "in recent years" | "since 2020" or "over the past five years" |
+| "some researchers" | Name them with citations |
+| "it is well known that" | Cite the source or remove |
+
+### Tense Usage
+| Section | Tense | Example |
+|---------|-------|---------|
+| Literature review (reporting findings) | Past | "Smith (2024) found that..." |
+| Literature review (ongoing state) | Present | "The theory posits that..." |
+| Methodology | Past | "Data were collected through..." |
+| Results | Past | "The analysis revealed..." |
+| Discussion (interpreting) | Present | "These findings suggest..." |
+| Conclusion (implications) | Present/Future | "This has implications for... / Future research should..." |
+
+## Chinese Academic Writing (zh-TW) Conventions
+
+### Register
+- Use written/formal language; avoid colloquial expressions
+- Prefer active voice (Chinese rarely uses passive)
+- Mainly short sentences; avoid overly long subordinate clauses
+- Use "this study" (ben yan jiu) rather than "we" (wo men)
+
+### Common Academic Expressions
+| English | Romanized Chinese |
+|------|------|
+| This study aims to | Ben Yan Jiu Zhi Zai |
+| The findings indicate | Yan Jiu Jie Guo Xian Shi |
+| It is worth noting | Zhi De Zhu Yi De Shi |
+| In conclusion | Zong Shang Suo Shu |
+| Based on the above analysis | Gen Ju Shang Shu Fen Xi |
+| Further research is needed | Wei Lai Yan Jiu Ke Jin Yi Bu Tan Tao |
+
+### Avoiding Translationese
+- Incorrect: "was found to be" (bei fa xian shi) → Correct: "Results show" (jie guo xian shi)
+- Incorrect: "This is because" (zhe shi yin wei) → Correct: "The reason lies in" (yuan yin zai yu)
+- Incorrect: "In the aspect of..." (zai...fang mian) → Correct: State directly
+- Incorrect: "It is worth being pointed out that" (zhi de bei zhi chu de shi) → Correct: "It is worth noting that" (zhi de zhu yi de shi)

package/vendor/ars/academic-paper/references/anti_leakage_protocol.md

@@ -0,0 +1,83 @@
+# Anti-Leakage Protocol (Knowledge Isolation)
+
+**Status**: v3.3
+**Used by**: `draft_writer_agent`, `report_compiler_agent`
+**Source**: Adapted from PaperOrchestra (Song et al., 2026, Appendix D.4)
+
+---
+
+## Purpose
+
+When the user provides comprehensive research materials (RQ Brief, Synthesis Report, Annotated Bibliography, experimental data), the writing agent should construct the paper **primarily from those materials**, not from LLM parametric memory. This prevents:
+
+1. **Methodology fabrication (Failure Mode 6)**: The LLM writes a plausible Methods section from training data rather than the user's actual procedure
+2. **Implicit knowledge leakage**: The LLM fills gaps with memorized content that may be inaccurate, outdated, or from a different context
+3. **Unintentional plagiarism**: The LLM reproduces near-verbatim passages from training data
+
+---
+
+## Protocol
+
+### When to activate
+
+Activate when ALL of the following are true:
+- The user has provided research materials via the pipeline handoff (RQ Brief + Synthesis Report + Annotated Bibliography)
+- The paper is in `full` or `revision` mode (not `plan` or `outline-only`)
+- The materials are substantive (not placeholder stubs)
+
+### When NOT to activate
+
+Do NOT activate when:
+- The user is in `plan` or `socratic` mode (exploratory — LLM knowledge is expected)
+- The materials are minimal (e.g., only a RQ Brief with no bibliography)
+- The user explicitly requests the LLM to supplement with its own knowledge
+
+### Prompt insertion
+
+When activated, prepend the following to the draft_writer_agent's working context:
+
+```
+## Knowledge Isolation Directive
+
+You are writing this paper based on the research materials provided in this session:
+- RQ Brief, Synthesis Report, and Annotated Bibliography (from deep-research)
+- Any additional materials the user has provided (experimental logs, datasets, prior drafts)
+
+Priority rules:
+1. PREFER session materials over your parametric knowledge for all factual claims
+2. Every claim in the paper MUST be traceable to a source in the Annotated Bibliography or user-provided data (#548 exception: an absence/novelty claim traces to the bibliography's documented Search Strategy — its bounded wording + nearest prior work — since no source can attest an absence; this is NOT a [MATERIAL GAP])
+3. If the materials do not cover a topic the outline requires, flag it as [MATERIAL GAP] rather than filling from memory
+4. Do NOT introduce references not present in the Annotated Bibliography unless explicitly asked by the user
+5. The Methods section must describe ONLY what is documented in the user's materials — do not infer or interpolate experimental procedures
+
+This is NOT a prohibition on using language skills or academic writing knowledge.
+You may use your knowledge of academic conventions, writing style, logical argumentation,
+and discipline norms. The restriction applies only to FACTUAL CONTENT — claims, citations,
+data, and methodology descriptions must come from session materials.
+```
+
+### [MATERIAL GAP] handling
+
+> **Tag vocabulary.** The `[MATERIAL GAP]`, `[WEAK EVIDENCE]`, `[GAP]` tags used throughout this protocol are canonically defined in [`shared/compliance_checkpoint_protocol.md#canonical-gap-tag-vocabulary`](../../shared/compliance_checkpoint_protocol.md#canonical-gap-tag-vocabulary). This section describes how the anti-leakage writing-time flag interacts with that vocabulary during manuscript production.
+
+When a `[MATERIAL GAP]` is flagged:
+1. The gap is surfaced at the next checkpoint
+2. The user can provide additional materials, or authorize LLM supplementation for that specific gap
+3. If supplemented: the gap section is tagged `[LLM-SUPPLEMENTED]` in the draft metadata for integrity review
+
+---
+
+## Relationship to existing checks
+
+| Check | What it catches | Anti-leakage adds |
+|-------|----------------|-------------------|
+| Integrity gate (Stage 2.5) | Fabricated citations post-hoc | Prevents fabrication at writing time |
+| Failure Mode 6 (Methodology fabrication) | Methods don't match actual procedures | Prevents LLM from inventing procedures |
+| Writing Quality Check | AI-typical phrasing patterns | Anti-leakage prevents AI-typical *content* (as opposed to style) |
+
+---
+
+## References
+
+- Song, Y. et al. (2026). PaperOrchestra. *arXiv:2604.05018*. Appendix D.4 (Anti-Leakage Prompt).
+- Lu, C. et al. (2026). Towards end-to-end automation of AI research. *Nature* 651, 914-919. — Mode 6 (Methodology fabrication).