Citation Scan

Count shared content words (nouns, verbs, adjectives — ignore stop words) between the claim and the matching source passage.

lexical_score = (shared_content_words / total_unique_content_words_in_both) × 100

Factor 2: N-gram Match (weight: 35%)

Check for shared phrases (n-grams of length 3-6 words). This catches structural copying that word-level analysis misses.

# Count matching 3-grams, 4-grams, 5-grams, 6-grams
# Weight longer matches more heavily:
ngram_score = (
    matching_3grams × 1.0 +
    matching_4grams × 2.0 +
    matching_5grams × 3.5 +
    matching_6grams × 5.0
) / max_possible_weighted_matches × 100

If any 8+ word sequence is identical → automatically set ngram_score = 100.

Factor 3: Structural Similarity (weight: 15%)

Does the claim preserve the sentence structure of the source?

• Same clause order + same logical flow = 100
• Reordered clauses but same structure = 60
• Different structure, same information = 30
• Completely restructured = 0

Factor 4: Semantic Equivalence (weight: 20%)

Does the claim convey the same meaning?

• Identical meaning, no added interpretation = 100
• Same core point with minor additions/context = 75
• Related but with user's own argument built on top = 40
• Only loosely connected conceptually = 15

Final Score

similarity = (lexical × 0.30) + (ngram × 0.35) + (structural × 0.15) + (semantic × 0.20)

Round to the nearest integer.

Score Thresholds

Search Procedure

Step 1: Search Notable Quotes

Read the "Notable Quotes & Citable Passages" section of each candidate source .md. Look for:

• Exact phrase matches (≥4 consecutive words)
• Key technical terms shared between claim and quote
• Same numerical results or statistics

Step 2: Search Paraphraseable Claims

Read the "Paraphraseable Claims" section. Compare:

• The user's claim against the Claim field (paraphrased version)
• The user's claim against the Original field (exact text)
• Use the Key terms field for fast relevance filtering

Step 3: Search Key Ideas and Results

Skim the "Key Ideas", "Key Results", and "Methodology" sections for conceptual matches.

Step 4: Deep PDF Scan (if no match found)

If no match is found in the .md files but the claim sounds like it could come from a source:

# Search the original PDF for key terms
pdftotext -layout [pdf_path] - | grep -i -C 5 "[distinctive term 1]"
pdftotext -layout [pdf_path] - | grep -i -C 5 "[distinctive term 2]"

Report the finding and note that the knowledge base summary may need updating.

Output Format

For each claim, produce this structured output:

## Claim [ID]: "[claim text]"

### Match Result
- **Best match source**: `[bibtex_key]` — "[Paper title]"
- **Matching passage**: "[exact text from source]"
- **Location**: Section [X], Page [Y]

### Similarity Breakdown
| Factor | Score | Weight | Weighted |
|---|---|---|---|
| Lexical overlap | [N]% | 0.30 | [N×0.30]% |
| N-gram match | [N]% | 0.35 | [N×0.35]% |
| Structural similarity | [N]% | 0.15 | [N×0.15]% |
| Semantic equivalence | [N]% | 0.20 | [N×0.20]% |
| **TOTAL** | | | **[SUM]%** |

### Classification
- **Similarity**: **[SUM]%** [EMOJI] [LABEL]
- **Overlapping phrases**: "[phrase 1]", "[phrase 2]", ...
- **What's too close**: [Explain which specific words/structures are borrowed]

### Recommendation
- **Action**: [Quote it / Rewrite + cite / Just cite / No action needed / Verify manually]
- **LaTeX suggestion**:
  ```latex
  [Corrected text with \cite{} command]

• If rewrite needed — suggested version:

  [Substantially rewritten text with \cite{} command]
  

If no match is found:

Claim [ID]: "[claim text]"

Match Result

• Best match source: NONE
• Classification: ❓ UNVERIFIABLE (0% match in knowledge base)
• Notes: [Possible explanation — common knowledge? Source not in KB?]
• Recommendation: Verify manually or index the source with the summarizer agent

## Rewrite Guidelines

When suggesting rewrites for 🔴 and 🟠 claims:

1. **Change sentence structure**: Active→passive or vice versa, reorder clauses, split/merge sentences
2. **Replace vocabulary**: Not just synonyms — rephrase the concept differently
3. **Integrate into argument**: Make the claim serve the user's narrative, not just restate the source
4. **Always include citation**: Even good paraphrases need `\cite{}`

Example:

Source: "Self-attention computes a weighted sum of all positions in a sequence, allowing the model to attend to relevant information regardless of distance."

🔴 Bad (92%): Self-attention computes a weighted sum across all sequence positions, enabling the model to attend to relevant information irrespective of distance.

🟡 OK (55%): The self-attention mechanism weights each token's contribution based on relevance rather than proximity, enabling long-range dependency capture \cite{vaswani2017_attention}.

✅ Good (28%): Unlike recurrent approaches that struggle with distant dependencies, attention-based models can directly relate any two positions in a sequence through learned relevance weights \cite{vaswani2017_attention}.

## LaTeX Citation Patterns

Suggest the appropriate pattern:

```latex
% Standard parenthetical
... as shown in prior work \cite{key}.

% Narrative (author is subject)
\textcite{key} demonstrated that...

% With page (for specific claims)
... as stated in \cite[p.~15]{key}.

% Direct quote (VERBATIM matches)
As \textcite[p.~7]{key} note, ``[exact quote]''

% Multiple sources supporting same claim
... consistent findings across studies \cite{key1, key2, key3}.

Special Cases

• Claim matches multiple sources: Report all matches with individual scores. Recommend citing the primary/original source, optionally citing others.
• Claim is a definition: If quoting a standard definition, suggest direct quote with \textcite. If the field has a standard definition, note it might be common knowledge.
• Claim includes numbers/results: Always needs citation. Match specific numbers against Key Results sections.
• Claim is in Slovak: Analyse the semantic content regardless of language. Match against English-language sources by meaning.