LJG-Xray-Paper: Paper X-Ray Machine

Core Positioning

Execution Steps

Step 1: Receive the Paper

Step 2: Execute Cognitive Extraction Algorithm

2.1 Denoising

• Ignore background introductions, polite remarks, and general known knowledge
• Skip lengthy Related Work sections (unless there are key comparisons)
• Filter out filler content written "for publication purposes"

2.2 Extraction

• Lock in the paper's core contribution (Delta)
• Identify the author's "aha moment"
• Find 1-2 key operations that determine success or failure

2.3 Critique

• Look for logical loopholes or boundary conditions
• Identify implicit assumptions
• Mark unresolved issues

Step 3: Structured Analysis

• One-sentence definition: What specific, difficult problem does this paper attempt to solve?
• Predecessor's dilemma: Why couldn't others solve it before this paper?

• Core intuition: The author's "aha" idea (in the plainest language possible)
• Key steps: List only 1-2 "stroke of genius" steps that determine success or failure

• vs SOTA: What specific improvements does it have compared to the current state-of-the-art?
• New puzzle piece: What specific new piece has it added to the human knowledge base?

• Implicit assumptions: Under what conditions can the author succeed?
• Unsolved mysteries: What problems did the paper not solve? What new problems did it bring?

• Napkin sketch: If you had to draw a diagram, what would you draw?
• Napkin formula: If you could only write one formula, what would it be?

Step 4: Generate Logical Structure Diagram

Step 5: Check System Environment and Select Output Format

1. Use Bash to execute

   command -v emacs >/dev/null 2>&1

   to check if emacs is available
2. Set the output format based on the check result:
   • If emacs exists (command returns 0): Use org-mode format
   • If emacs does not exist (command returns non-0): Use markdown format

Step 6: Generate Report

• Precise, concise, and clear text
• Use natural paragraphs, not tables
• ASCII Art: Only use pure ASCII basic symbols, no Unicode

Org-mode Template (when emacs exists)

#+title:      xray-{short-title}
#+date:       [{YYYY-MM-DD Day HH:MM}]
#+filetags:   :read:xray:paper:
#+identifier: {YYYYMMDDTHHMMSS}
#+source:     {paper link}
#+authors:    {authors}
#+venue:      {conference/journal}

* NAPKIN FORMULA

#+begin_example
+----------------------------------------------------------+
|                                                          |
|   {napkin formula}                                        |
|                                                          |
+----------------------------------------------------------+
#+end_example

{one-sentence explanation of the formula}

* PROBLEM

**Pain Point Definition**: {one-sentence problem definition}

**Predecessor's Dilemma**: Why couldn't others solve it before?

* INSIGHT

**Core Intuition**: The author's "aha" idea (in plainest language)

**Key Steps**:
1. {stroke of genius 1}
2. {stroke of genius 2}

* DELTA

**vs SOTA**: Specific improvements compared to current state-of-the-art

**New Puzzle Piece**: What specific new piece was added to the human knowledge base?

* CRITIQUE

**Implicit Assumptions**:
- {assumption 1}
- {assumption 2}

**Unsolved Mysteries**:
- {unresolved issue 1}
- {unresolved issue 2}

* LOGIC FLOW

#+begin_example
{pure ASCII logical structure diagram: Problem --> Insight --> Method --> Result}
#+end_example

* NAPKIN SKETCH

#+begin_example
{napkin sketch: core concept drawn with ASCII}
#+end_example

Markdown Template (when emacs does not exist)

# xray-{short-title}

**Date**: {YYYY-MM-DD Day HH:MM}
**Source**: {paper link}
**Authors**: {authors}
**Venue**: {conference/journal}
**Tags**: read, xray, paper
**Identifier**: {YYYYMMDDTHHMMSS}

## NAPKIN FORMULA

{one-sentence explanation of the formula}

## PROBLEM

**Pain Point Definition**: {one-sentence problem definition}

**Predecessor's Dilemma**: Why couldn't others solve it before?

## INSIGHT

**Core Intuition**: The author's "aha" idea (in plainest language)

**Key Steps**:
1. {stroke of genius 1}
2. {stroke of genius 2}

## DELTA

**vs SOTA**: Specific improvements compared to current state-of-the-art

**New Puzzle Piece**: What specific new piece was added to the human knowledge base?

## CRITIQUE

**Implicit Assumptions**:
- {assumption 1}
- {assumption 2}

**Unsolved Mysteries**:
- {unresolved issue 1}
- {unresolved issue 2}

## LOGIC FLOW

## NAPKIN SKETCH

undefined

Step 7: Save and Open

1. Generate timestamp: Use Bash to execute

   date +%Y%m%dT%H%M%S

   to get the current time
2. Determine the file extension based on the check result from Step 5:
   • If using org-mode format: extension is

     .org

   • If using markdown format: extension is

     .md

3. File name format (Denote specification):

   {timestamp}--xray-{short-title}__read.{extension}

   • Short title: Take the first 3-5 keywords from the paper title, lowercase, connected with hyphens
   • Example (org):

     20260207T171500--xray-dflash-block-diffusion__read.org

   • Example (markdown):

     20260207T171500--xray-dflash-block-diffusion__read.md

4. Save path:

   ~/Documents/notes/{file name}

5. Execute via Bash:

   open ~/Documents/notes/{file name}

Output Quality Standards

• High Density: Use lists and keywords, no long paragraphs
• Plain Language: Explain complex concepts in the simplest language possible
• Critical: Must point out at least one implicit assumption or unresolved issue
• ASCII Art: Only use pure ASCII basic symbols, no Unicode
• Napkin Sketch/Formula: Must be understandable at a glance