Skip to content

PERSONAL_STYLE_PROFILE.md

Latest commit

 

History

History
91 lines (64 loc) · 5.73 KB

PERSONAL_STYLE_PROFILE.md

File metadata and controls

91 lines (64 loc) · 5.73 KB
Raw

Current Lu-Xing Yang IEEE Transactions Style Profile

Version: 0.4.0 Status: expanded-partial-calibration

Evidence base

The research doctrine now uses a verified 73-paper full-text working corpus: 42 owner-provided papers plus 31 nonduplicate open-access journal articles discovered through the owner's publication identity and verified to an actual PDF or complete publisher HTML. The 42-paper local corpus contains 17 IEEE Transactions papers, a 20-paper first/corresponding-author core-voice subset, and 19 papers from 2024 onward. The selected 18-row Transactions manifest has 17 private-full-text-covered priority rows and one explicitly deprioritized different-subfield row.

The expanded set is still not a verified complete publication universe. Sentence-level preferences remain driven by the role-weighted local corpus, especially first/corresponding-author or documented original-draft papers. The 31 open-access additions strengthen research architecture, mathematical/algorithmic quality gates, method evolution, and consistency checks; they do not automatically become personal sentence-voice evidence.

Calibration limits

  • The 20-paper core-voice tier is a weighting set, not proof that every sentence or section was personally drafted by the owner. Corresponding-author status remains a signal, and many contribution roles are still unknown.
  • The expanded 73-paper deep-reading set is architecture and correction evidence. It does not change the 20-paper personal-voice core or the attachment-derived quantitative sentence targets.
  • Abstract-level architecture and sentence rhythm are better supported than paragraph-level, section-specific, collaborator-specific, or punctuation-level microstyle.
  • TIFS has the strongest venue-specific Tier-A signal; TDSC, TSMC, and TCSS personal-voice samples remain too thin for strong author-by-venue claims.
  • Topology-first simulation, RL/MARL/MPC, and PINN/PIDL include emerging or repository-derived capability. Do not describe them as equally mature publication-corpus directions.
  • Use references/corpus/local_corpus_metrics.json as the quantitative source of truth and treat prose summaries as rounded interpretations.
  • Use references/FULLTEXT_CORPUS_DERIVED_DOCTRINE.md for cross-paper model/theory/code/evidence checks and references/corpus/open_access_fulltext_manifest.csv for the public 31-paper expansion boundary.

One-line signature

Write from an operational cyber/network decision problem, separate the mechanism and intervention gaps, formalize a named model/problem, derive the solution conditions, implement a verifiable solver, and close with comparative evidence and a bounded cost/timing/topology/security implication.

Current structure

Introduction
  stakes → mechanism difficulty → literature taxonomy → dual gap
  → proposed approach → model/theory/algorithm/evidence contributions

System model / problem formulation
  entities → states/units → topology → timing → actions/information
  → objective/payoff → admissibility → solution concept

Theory
  assumptions → theorem ladder → necessary/sufficient status
  → proof/dependency → numerical meaning

Algorithm
  inputs → initialization → updates → projection/reset
  → stopping residual → complexity → failure behavior

Evidence
  sanity/independent check → matched baselines
  → topology/parameter/seed coverage → sensitivity/ablation
  → uncertainty/failures → scalability → bounded implications

Quantitative soft targets

  • Abstract: 190–240 words, normally 8–11 sentences.
  • Mean sentence length: approximately 19–26 words.
  • Review sentences above 42–45 words.
  • Manually inspect every sentence above 60 words.
  • Contribution list: normally 2–4 items in model → theory → algorithm → evidence order.

Preferred prose behavior

  • Use active we for research actions.
  • Name the formal object: model, problem, theorem, optimality system, algorithm, estimator, policy.
  • Put the purpose or claim at the start of a paragraph.
  • Use equations after the entities, units, timing, and meaning are clear.
  • Narrate results as metric → magnitude/uncertainty → mechanism → bounded implication.
  • Use indicate or suggest when evidence is conditional; reserve establish for formal or sufficiently decisive evidence.
  • State the exact model, topology, parameter, information, budget, and scenario boundary.

Phrases to use functionally, not mechanically

However, To address ..., Based on ..., On this basis, Next, Furthermore, Finally, formulate, derive, establish, develop, optimality system, iterative algorithm, comparative experiments.

Do not repeat the same connector in adjacent sentences or more than roughly three times per 500 words.

Corrections to historical habits

Do not imitate grammar defects, vague usefulness claims, thereby evidencing, unsupported first/novel/innovative, or a claim of optimality/equilibrium based on one numerical run.

Experiment fallback

When real propagation or attack–defense trajectories cannot be obtained:

  1. acquire a semantically compatible real topology when possible;
  2. call the study real-topology simulation, not real-world validation;
  3. otherwise use a matched, multi-family synthetic topology suite;
  4. generate transparent parameter, initial-state, and attack-seed scenarios;
  5. verify the node-level model with an independent solver or stochastic ensemble;
  6. use matched baselines, multiple graph/parameter seeds, holdouts, uncertainty, and failure logs;
  7. scope every conclusion to the declared model/topology/parameter assumptions.

See references/ATTACHMENT_CORPUS_ANALYSIS.md, references/FULLTEXT_CORPUS_DERIVED_DOCTRINE.md, and references/TOPOLOGY_FIRST_EXPERIMENT_PROTOCOL.md.