Document Metadata Reference

---

Decision‑Making Insights

• Structural consistency: Verify total_sections and review chapters to ensure logical chapter breaks.
• Narrative flow: Chart aggregated_sentiment and top_topics to understand theme and tone progression.
• Readability targets: Compare avg_readability to your intended audience’s reading level.
• Genre alignment: Use genre and global_top_keywords to posit~~ion your manuscript in the right market.
• Marketing hooks: Leverage global_top_keywords and top_topics for blurbs, tags, and promotional copy.

How each analytics tile can guide your edits

UI tile / metric block	What it reveals	How to use it to polish the draft
Bag of Words / Phrases (top single-words & bigrams)	Over-used lexis and recurring images.	Replace the most common verbs/nouns with fresher synonyms, trim filler (“back”, “like”), or lean into a deliberate motif if repetition is intentional.
Sentiment gauge + bubbles	Overall emotional temperature (neg/neu/pos) and polarity balance.	If the gauge is stuck on the negative side, seed in lighter beats or hopeful reactions to give readers moments to breathe. Conversely, dial up tension if the story feels too neutral.
Word Count / Est. Pages / Read-Time	Raw size and pacing expectations.	Compare to genre norms: if your YA chapter runs 9 000 words (≈30 min), consider tightening scenes or splitting chapters for younger audiences.
Readability (Flesch–Kincaid, Fog, SMOG)	Sentence complexity & estimated U.S. grade level. • Flesch–Kincaid = grade level based on sentence length and syllables. • Gunning Fog = grade level based on sentence length and complex words. • SMOG = grade level based solely on polysyllabic words.	Aim for 4–6 for middle-grade, 6–9 for YA, and 10+ only for academic or literary prose. Lower scores mean more accessible reading. Adjust by simplifying sentence structure and vocabulary.
Lexical metrics (TTR, Hapax, Avg Word Len)	Vocabulary richness and variety. • TTR ≥ 0.5 = dense variety; < 0.25 = repetitive. • Hapax shows one-off words count. • Longer avg word-len hints at more advanced diction.	Boost TTR/Hapax by pruning echoes and adding sensory specifics; lower them when a simpler voice is desired.
Dialog Ratio	% of dialogue lines vs narrative exposition.	Heavy narration (< 30 % dialogue) may feel slow—inject conversations. If > 60 %, add grounding action and interiority so scenes don’t float.
Top Topics (add-on)	Theme clusters extracted via LDA/BERTopic.	Verify that dominant topics match your intended themes; amplify or trim scenes to keep focus.
NER / Beats / Tropes / AI metric (add-on)	Named-entity consistency, plot-beat coverage, trope tagging, and overall “engagement” score.	Catch name misspellings, ensure required beats occur at genre-expected positions, and target trope usage for marketing blurbs.
Sentiment Trend (sparkline per chapter)	Emotional arc across the manuscript.	Check for flatlines: long stretches of identical sentiment can dull tension. Vary peaks/valleys to sustain reader engagement.
Word Trends (per-chapter counts)	Pace & scene density.	Sudden spikes may signal info-dumps; valleys might indicate scenes too sparse—rebalance for rhythm.

Quick example (excerpt)

{
  "readability": { "flesch_kincaid": 4.8, "gunning_fog": 7.2, "smog": 8.1 },
  "lexical_metrics": { "ttr": 0.31, "hapax": 523, "avg_word_length": 4.43 },
  "dialog_ratio": 0.54
}

Interpretation:

• F-K 4.8 → approachable for middle-grade readers.
• TTR 0.31 → healthy variety; no urgent action.
• Dialog 54 % → dialogue-driven; sprinkle more scene texture if pacing feels rushed.

Interpretation — Document-level Word & Sentiment Trends

• Word sparkline
  Tall spikes mark chapters > 2 SD above mean word count—candidates for trimming or splitting.
  Deep valleys highlight under-developed scenes that may need fleshing out.

• Average line (grey)
  Reveals drift: if late chapters all hover well above the line, pacing is inflating—rebalance earlier sections.

• Sentiment bars

  • Green = % positive sentences* Red = % negative.
    Large red stacks warn of sustained gloom; aim for varied stacks to keep emotional movement.

• Compound line
  –100 → +100 single-number polarity summary per chapter.
  • While the bars show how much of each polarity is present, compound blends them into one value (≈ weighted pos – neg). A chapter with modest bars can still have a strong +/– compound if one side dominates.
  • Use it to spot overall vibe at a glance, then inspect bar mix for nuance.

• Smoothed net curve
  A gentle roller-coaster = healthy emotional rhythm. A long plateau (± 5 %) signals monotony—add a reversal or comic relief.

info

Advanced analytics—Top Topics, Named-Entity consistency, Trope & Beat detection, and AI-powered engagement scoring—are available as an add-on at checkout. Unlock them when you need deeper structural or marketing insights.

Use these panels as a revision checklist: tackle readability and repetition first (highest return for effort), refine sentiment flow next, and finish with advanced add-ons for final polish.

Aggregated insights at the manuscript level

Field	Type	Description
total_sections	usize	Number of sections/chapters analyzed.
total_words	usize	Sum of all word counts.
total_paragraphs	usize	Sum of all paragraph counts.
total_chars	usize	Sum of all character counts.
avg_read_time_min	f64	Average reading time per section (in minutes).
global_term_freqs	HashMap<String, usize>	Top N words by frequency across the document.
global_bigrams	HashMap<String, usize>	Top N two-word phrases across the document.
global_top_keywords	HashMap<String, f64>	Top N TF–IDF keywords for the entire manuscript.
aggregated_sentiment	Sentiment	Mean sentiment (neg/neu/pos/compound) averaged over sections.
top_topics	Vec<String>	Most frequent topics across all sections.
genre	Option<String>	Majority-genre label based on section genres.
avg_sentence_length	f64	Mean sentence length across sections.
avg_ttr	f64	Mean type-token ratio across sections.
avg_hapax	f64	Mean hapax legomena count per section.
avg_word_length	f64	Mean average word length per section.
avg_readability	ReadabilityMetrics	Mean readability scores (Flesch–Kincaid, Gunning Fog, SMOG) across sections.
avg_dialog_ratio	f64	Mean dialogue ratio (proportion of words in dialogue) averaged over sections.
author	Option<String>	First detected author (if any) from the opening sections.
chapters	Vec<ChapterInfo>	List of each chapter’s info: ChapterInfo { title, subtitle, words, sentiment, dialog_ratio }.