CQ.AI Similarity Search Algorithms

Overview

Similarity search compares text by converting each entry into a numeric vector. CQ converts each text entry into a 512‑dimensional vector by using the Universal Sentence Encoder Multilingual module.

The module supports 16 languages and compares multilingual content using the same vector format. It accepts variable‑length text, including sentences, phrases, and short paragraphs. For details, see https://tfhub.dev/google/universal-sentence-encoder-multilingual-large/3

Preprocessing for Semantic Similarity 

Note: For details, see the CQ AI Settings page for the Similarity Search algorithm.

Additional Resources

ComplianceQuest

• CQ.AI k-Nearest Neighbors Algorithm (KNN)

• CQ.AI Decay Algorithms