Technical Terms

Semantic optimization involves structuring content with clear relationships, entities, and context that AI models can understand and correctly interpret.

A Knowledge Graph is how AI models organize and connect information about entities (brands, products, concepts) and their relationships.

RAG (Retrieval-Augmented Generation) is a technique where AI models retrieve relevant documents or data before generating responses.

Also: Retrieval-Augmented Generation

AI scraping is a methodology where tools automatically query AI models and collect responses to track brand mentions, sentiment, and positioning.

Also: Response scraping, AI monitoring

Gender bias in AI refers to the documented phenomenon where AI models recommend different brands depending on how a query is gender-framed.

Also: Gender-locked brands, AI gender categorization

An ontology cluster is a collection of semantically related topics, keywords, and concepts that AI models group together when understanding your brand.

Platform preference patterns describe how different AI platforms have distinct biases in how they recommend brands.

Also: AI platform behavior, Model bias patterns

Category gatekeeping occurs when AI has learned to associate brands only with specific categories, excluding them from adjacent or broader categories.

Also: AI category bias, Systematic exclusion

Embeddings are numerical representations of text, images, or other data that capture semantic meaning in a format AI systems can process mathematically.

Vector search is the technology AI systems use to find semantically similar content even when there are no exact keyword matches between a query and a document.

Semantic search is an approach to information retrieval that understands the meaning and intent behind a query rather than simply matching keywords.

A context window is the maximum amount of text a large language model can process in a single interaction.

A token is the smallest unit of text that an AI model processes. On average, one token equals approximately three-quarters of a word in English, meaning a 1,000-word article contains roughly 1,333 tokens.

Temperature is an AI model setting that controls the randomness and creativity of generated responses.

Fine-tuning is the process of training an existing large language model on a specialized dataset to improve its performance in a specific domain or for a particular task.

RLHF (Reinforcement Learning from Human Feedback) is the training technique used to teach AI models which responses humans prefer.

Also: Reinforcement Learning from Human Feedback

Schema markup for AI refers to the use of structured data code (typically JSON-LD) on your website to help AI systems identify and understand your brand, products, expertise, and relationships.

Also: Structured Data for AI, AI-Optimized Schema

Entity recognition (also known as Named Entity Recognition or NER) is the AI capability of identifying and distinguishing specific brands, products, people, organizations, and concepts within text.

Also: Named Entity Recognition, NER