Compare text

The Cortical.io Retina performs the following operations:

1. Parses the input text.
2. Detects the input language of each of the two pieces of text if the text is at least 10 words and at least 40 characters.
3. Numerically encodes the meaning of the two pieces of text as two semantic fingerprints.
4. Displays the semantic fingerprints as square grids, where each blue dot on a grid contains part of the meaning of the text; a third grid, between the other two, indicates with orange dots where the left and right grids overlap.
5. Quantifies the overlap in meaning between the two pieces of text as a percentage, based on the number of overlapping points between the items to compare.

The Cortical.io Retina technology is accessed via a REST API. To detect languages and to create, compare, and display semantic fingerprints, you can include the following API calls in your application:

• /text/detect language endpoint: Can identify more than 50 languages
• /text endpoint: Creates a semantic fingerprint for inputted text
• /compare endpoint: Calculates the similarity of meaning between two inputs
• /image/compare endpoint: Generates a fingerprint comparison image (left image, overlap, and right image)

To begin to integrate Cortical.io services into your application, register for a free API key.

Comparing text or web pages

To compare text or web pages, follow the instructions on the screen.

The meaning of the two inputs is numerically encoded as two semantic fingerprints, which are graphically displayed as two square grids. Each blue dot on a grid contains part of the meaning of the text. A third grid, between the other two, indicates with orange dots where the left and right grids overlap.

The overlap in meaning between the two inputs is quantified as a percentage, based on cosine similarity.

Exploring fingerprints

Every dot on a semantic fingerprint represents a collection of terms that are associated with one another. You can view some of the terms that are located at any of the colored dots on the grid by hovering over that dot. Dots that are close to one another on the grid are also close in meaning. Together, all colored dots on the grid represent the meaning of the inputted text or web page.

Use cases

• Meaning-based searches
• Content filtering
• Content classification
• Profile matching