Evidenzia

Explore Scripture by Evidence

A computational lens for theological inquiry. We map the conceptual architecture of sacred texts using meaning fingerprints, allowing for objective structural discovery.

The Problem

Theological debates often rely on selective verse use. Each tradition emphasizes certain passages and downplays others. Claims like “the Bible clearly teaches X” or “there are more verses supporting Y” are rarely tested against the full text. Evidenzia exists to make those claims testable.

Tradition A Interpretation

“The context implies a physical state of rest...”

Tradition B Interpretation

“Historical usage suggests a metaphysical departure...”

How Evidenzia Works

Verse

Raw KJV Text

Tag

Keyword Match

Meaning

Semantic Match

Classify

Multi-Signal Vote

Challenge

Counter-Argument

Visualize

Relational Map

31,102
KJV Verses Indexed
Semantic
Meaning-Based Discovery
4-Layer
Depth of Analysis

Meaning Fingerprints

Every verse is converted into a 768-value numerical representation of its meaning. Think of it as coordinates in a high-dimensional space where verses with similar meanings cluster together.

The system does not assign theology. It only measures similarity in meaning. Verses that discuss similar ideas are close together. Verses on unrelated topics are far apart.

Similarity Score98.4%
Conceptual Delta0.021
Vector Projection
Closest Matches
Eccl. 9:594.2%
Psalm 13:391.8%
Job 14:1289.5%

See It in Action

Pick a verse and see its meaning fingerprint, then see which other verses have the closest fingerprints.

Genesis 1:1

In the beginning God created the heaven and the earth.

Step 1AI reads the verse and assigns 768 numbers

Each number represents one aspect of the verse’s meaning. Together, these 768 numbers form a unique “fingerprint” for this verse. Here are the first 16:

... 752 more
Dimension 1Dimension 16
[-0.025, 0.016, -0.002, -0.044, 0.004, 0.041, 0.000, -0.024, -0.005, -0.006, -0.023, 0.010, -0.013, 0.003, 0.114, -0.006, ...]
Step 2Find verses with similar fingerprints

We compare this fingerprint against all 31,102 verses and find the closest matches. No theology involved, just measuring which verses have the most similar numbers.

74%
match
Genesis 2:1

Thus the heavens and the earth were finished, and all the host of them.

74%
match
Genesis 2:4

These are the generations of the heavens and of the earth when they were created...

68%
match
Genesis 5:1

This is the book of the generations of Adam. In the day that God created man...

67%
match
Genesis 1:27

So God created man in his own image, in the image of God created he him...

How Verses Are Identified

Each theme is built through four layers of analysis. No single layer decides — they vote together, and the final layer challenges every result.

01. Keyword Matching

Each theme begins with clearly defined words and phrases. For example, “Death as Sleep” includes terms like “sleepeth,” “fallen asleep,” and “sleep in the dust.”

This layer is fully transparent. You can see exactly which words triggered the tag.

“The dead know not any thing...”

02. Meaning-Based Expansion

Using the embedding model, the system finds verses that are close in meaning to those identified by keywords, even if they use different language.

These results are based on mathematical similarity, not interpretation.

03. Multi-Signal Classification

Multiple independent signals vote on whether a verse belongs to a theme: keyword presence, meaning similarity, linguistic structure (via spaCy NLP), and contextual domain words. A probability model (Snorkel) learns which signals are reliable and how they correlate, then produces a single confidence score.

Hard exclusion rules vote against inclusion — genealogies, ritual durations, and non-prophetic time references are automatically filtered. Only verses that pass the probability threshold proceed to the final layer.

Keyword
Semantic
Pattern
→ Probability Model →
Score

04. Counter-Argument Evaluation

Each tagged verse is evaluated: how strongly does it connect to the theme? Then the system generates the strongest argument that it does not belong. This produces a relevance score and ensures every tag carries its own counter-evidence. A high relevance score means the verse is deeply connected to the conversation — not that it conclusively proves the position.

Weak or ambiguous connections are reduced in strength or removed.

Supports
Challenges

Reading the Analytics

The percentage you see on each verse is a relevance score, not a truth claim. It measures how strongly a verse connects to a theme through language, meaning, and context. A verse at 100% relevance means the system is highly confident the verse belongs in the conversation about that theme — not that it definitively proves the position. Every tagged verse carries a counter-argument showing the strongest case against that interpretation. The data maps connections. The theology is yours.

Relevance % and Source Colors

75%
42%
Keyword Match
Semantic Match

Delta Analysis

Sleep
69
Δ 67
Conscious
2

Distribution across the Old and New Testament canon.

Neutrality Safeguards

Keywords are defined by vocabulary, not doctrine.
Meaning similarity is based on mathematical distance.
Every verse is paired with a counter-argument.
All tags record how they were generated.
No tradition-specific framework is applied.
The tool shows structure. Theology is yours.

Etymological Precision

Every term is tied to its Strong’s concordance identifier. The system looks at the underlying semantic structure of the original manuscripts, not just English translations. This helps clarify distinctions that translations often flatten.

שְׁאוֹל
SheolH7585
Primary Meaning

The underworld, state of the dead, grave.

KJV Rendering Frequency
Grave (31x)Hell (31x)Pit (3x)

“Evidenzia does not interpret Scripture for you. It reveals the structure of the text so you can evaluate it yourself.”

Explore a Concept