issues/14-progress.md

Phase 14 Progress Report

Phase 14 Goals

"Narrative Tapestry — From Spindle to Woven Cloth"

Phase 14 extends the semantic navigation system from poems to long-form narrative. By segmenting books into passages and applying the same embedding/similarity infrastructure, readers can navigate stories through semantic relationship rather than purely linear sequence.

The truth is, the story is what's important, and it's often best told with chains of words strung together into a long message starting from the beginning and going until the end, with breakpoints interspersed throughout so you don't lose track of the thought. However, I can't help but wonder if we take that spindle of thread and we wove it into a tapestry, guided by modern technology (LLMs) and assembling something greater than us and ourselves.

From Previous Phases

Complete embedding infrastructure (Ollama + EmbeddingGemma)
Similarity matrix computation (triangular storage, GPU acceleration)
Similar/Different/Chronological navigation paradigm
Static HTML generation pipeline
Word cloud and semantic color systems

Phase 14 Objectives

Adapt the system to process book-length texts
Implement intelligent passage segmentation (paragraph-based, with bundling)
Generate navigable book representations with Similar/Different/Chronological pages
Test with diverse text types (fables, philosophy, poetry, fiction)
Explore what new meanings emerge when linear narrative becomes navigable tapestry

Phase 14 Issues

Active Issues

Issue	Description	Status	Priority
14-001	Book Tapestry Navigation System (Vision)	Open	High

Sub-Issues (To Be Created)

Sub-Issue	Description	Status
14-001a	Design book input format and metadata schema	Planned
14-001b	Implement passage segmentation algorithm	Planned
14-001c	Adapt embedding pipeline for book passages	Planned
14-001d	Generate book-specific HTML templates	Planned
14-001e	Test with sample texts	Planned

Completed Issues

None yet.

Key Concepts

The Spindle and the Tapestry

Traditional reading treats a book as a spindle of thread — you follow it from beginning to end. This system weaves that thread into a tapestry where:

Similar navigation: Find passages that resonate with what you just read
Different navigation: Seek contrast and surprise
Chronological navigation: Return to the original linear flow

The author's words remain unchanged. Only the reader's path through them expands.

Passage Segmentation

Books are divided into passages (typically paragraphs), with small passages bundled together:

Paragraph 1 (50 chars)  ─┐
Paragraph 2 (30 chars)   ├─→ Passage 1 (180 chars)
Paragraph 3 (100 chars) ─┘

Paragraph 4 (500 chars) ───→ Passage 2 (500 chars)

Paragraph 5 (80 chars)  ─┐
Paragraph 6 (200 chars) ─┴─→ Passage 3 (280 chars)

This ensures each passage has enough semantic content for meaningful embedding.

Suitable Texts

Well-suited:

Episodic narratives (fables, mythology)
Non-linear structures (experimental fiction)
Philosophical texts (aphorisms, meditations)
Poetry collections (already proven)
Sacred texts (designed for cross-reference)

Less suited (but still interesting):

Tightly plotted mysteries (spoiler structure)
Character-driven novels (emotional arcs)

Target Sample Texts

Public domain candidates for testing:

Aesop's Fables — Short, episodic, thematically rich
Meditations by Marcus Aurelius — Aphoristic, non-linear by nature
Grimm's Fairy Tales — Episodic, thematic patterns
The Prophet by Kahlil Gibran — Poetic prose, thematic chapters

Completion Criteria

[ ] Book input format and metadata schema defined
[ ] Passage segmentation algorithm implemented
[ ] Embedding pipeline adapted for book-length texts
[ ] HTML templates created for book navigation
[ ] At least one sample book successfully processed
[ ] Similar/Different/Chronological navigation functional

Phase Status: OPEN

Created: 2026-01-28

Cross-Phase Dependencies

Depends on:

Phase 1-8: Embedding infrastructure, HTML generation, similarity computation
Phase 9: GPU acceleration (for large books)

Enables:

New forms of literary exploration
Multi-book corpus navigation (future)
Comparative literature tools