The generative AI world moves at breakneck pace, so there isn’t always time to really sink your teeth into important topics before the next shiny item pops up in the news cycle. But there are a few core concepts that are critical to understand when it comes to GenAI.
To help refresh your knowledge, we’ve assembled a set of in-depth guides across a variety of key topics around vector search. They’re free, you’re not required to leave your email address or fill out an annoying form, and they’ve proved very popular, in part because they appeal to a spectrum of interests—from developers to tech leaders.
What is a vector database?
A vector database is the backbone of generative AI. It’s a specialized storage system designed to efficiently handle and query high-dimensional vector data, which is commonly used in AI and machine learning applications for fast and accurate data retrieval.
Read the vector database guide
What is vector search?
Vector databases enable vector search, which is a method of data retrieval that uses mathematical vectors to represent and efficiently search through complex, unstructured data. It helps you discover linked concepts in search responses, rather than solely focusing on keywords.
What is vector indexing?
A vector index is a data structure used in computer science and information retrieval to efficiently store and retrieve high-dimensional vector data, enabling fast similarity searches (also known as vector search) and nearest neighbor queries (more on this below).
Read the vector indexing guide
What is cosine similarity?
Cosine similarity is a mathematical measurement of how similar two vectors are. In essence, it helps you understand the relationship between two elements by looking at the "direction" they are pointing in, rather than just comparing them based on their individual values.
It’s often used in vector search engines to find the most relevant records to a given query, making search processes more efficient and precise.
Read the cosine similarity guide
What is the nearest neighbor algorithm?
You can think of the nearest neighbor algorithm as a subset of vector search, as it’s a specific technique used to facilitate vector search. It’s another piece of the puzzle that produces highly relevant results to prompts.
A nearest neighbor algorithm plots all vectors in a multi-dimensional space and uses each of the points to find a neighboring point that is nearest. There’s a host of different types of nearest neighbor algorithms you can use, depending on the precision required by your app.
By the way, this guide is one of our favorites, as it explains the nearest neighbor algorithm in the context of charcuterie boards and cheese. 🧀
Read the nearest neighbor algorithm guide
What are hierarchical navigable small worlds?
This one’s a mouthful, but hierarchical navigable small worlds (HNSW) are important because they provide an efficient way to navigate and search through complex, high-dimensional data.
By organizing data hierarchically and enabling navigable shortcuts, HNSW significantly reduces the time and computational resources needed for these searches, making it a crucial tool in handling large datasets in machine learning and AI.
Read the hierarchical navigable small worlds guide
Conclusion
Vector databases and vector search are critical concepts to understand, and critical foundations for any organization that is building GenAI applications.
Astra DB is a leading vector database, but it’s set apart from many rivals by the fact that it’s a highly performant, fast, powerful, and reliable database for all the data that’s required for AI applications – operational data and vectors. You can give it a try for free.
For a look at other important AI topics, including AI agents, embeddings, and model context protocol (MCP), check out our AI Learning Center.
