{"id":15353,"date":"2024-02-20T10:35:25","date_gmt":"2024-02-20T18:35:25","guid":{"rendered":"https:\/\/www.couchbase.com\/blog\/?p=15353"},"modified":"2025-06-13T22:42:02","modified_gmt":"2025-06-14T05:42:02","slug":"what-are-vector-embeddings","status":"publish","type":"post","link":"https:\/\/www.couchbase.com\/blog\/what-are-vector-embeddings\/","title":{"rendered":"What are Vector Embeddings?"},"content":{"rendered":"

Vector embeddings are a critical component in machine learning that convert “high-dimensional” information, such as text or images, into a structured vector space. This process enables the ability to process and identify related data more effectively by representing it as numerical vectors. In this post, you’ll learn how to create vector embeddings, their types, and their deployment in various use cases.<\/span><\/p>\n

Vector Embeddings Explained<\/h2>\n

Vector embeddings are like translating information we understand into something a computer understands. Imagine you\u2019re trying to explain the concept of “Valentine\u2019s Day” to a computer. Since computers don\u2019t understand concepts like holidays, romance, and the cultural context the way we do, we have to translate them into something they DO understand: numbers. That\u2019s what vector embeddings do. They represent words, pictures, or any kind of data in a list of numbers that represent what those words or images are all about.<\/span><\/p>\n

For example, with words, if “cat” and “kitten” are similar, when processed through a (large) language model, their number lists (i.e., vectors) will be pretty close together. It\u2019s not just about words, though. You can do the same thing with photos or other types of media. So, if you have a bunch of pictures of pets, vector embeddings help a computer see which ones are similar, even if it doesn\u2019t “know” what a cat is.<\/span><\/p>\n

Let\u2019s say we\u2019re turning the words “Valentine\u2019s Day” into a vector. The string “Valentine\u2019s Day” would be given to some model, typically an <\/span>LLM (large language model)<\/span><\/a>, which would produce an array of numbers to be stored alongside the words.<\/span><\/p>\n

{\r\n\u00a0 \"word\": \"Valentine's Day\",\r\n\u00a0 \"vector\": [0.12, 0.75, -0.33, 0.85, 0.21, ...etc...]\r\n}<\/pre>\n
Vectors are very long and complex. For instance, <\/span>OpenAI\u2019s vector size<\/span><\/a> is typically 1536, which means each embedding is an array of 1536 floating point numbers.<\/span><\/p>\n
 <\/p>\n
\"Produce<\/a><\/p>\n
 <\/p>\n
By itself, this data doesn\u2019t really mean much: it\u2019s all about finding other embeddings that are <\/span>close<\/span><\/i>.<\/span><\/p>\n
 <\/p>\n
\"Produce<\/a><\/span><\/p>\n
In this diagram, a nearest neighbor algorithm can find data with vectors <\/span>close<\/span><\/i> to the vectorized query. These results are returned in a list (ordered by their proximity).<\/span><\/p>\n
Types of Vector Embeddings<\/h2>\n
There are several types of embeddings, each with its unique way of understanding and representing data. Here\u2019s a rundown of the main types you might come across:<\/span><\/p>\n
Word Embeddings<\/b>: Word embeddings translate single words into vectors, capturing the essence of their meaning. Popular models like Word2Vec, GloVe, and FastText are used to create these embeddings. These can help to show the relationship between words, like understanding that “king” and “queen” are related in the same way as “man” and “woman.”<\/span><\/p>\n
Here\u2019s an example of Word2Vec in action:<\/span><\/p>\n
from gensim.models import Word2Vec\r\n\r\nsentences = [\r\n\u00a0 \u00a0 \"Couchbase is a distributed NoSQL database.\",\r\n\u00a0 \u00a0 \"Couchbase Capella provides flexibility and scalability.\",\r\n\u00a0 \u00a0 \"Couchbase supports SQL++ for querying JSON documents.\",\r\n\u00a0 \u00a0 \"Couchbase Mobile extends the database to the edge.\",\r\n\u00a0 \u00a0 \"Couchbase has a built-in Full Text Search Engine\"\r\n]\r\n\r\n# Preprocess the sentences: tokenize and lower case\r\nprocessed_sentences = [sentence.lower().split() for sentence in sentences]\r\n\r\n# Train the Word2Vec model\r\nmodel = Word2Vec(sentences=processed_sentences, vector_size=100, window=5, min_count=1, workers=4)\r\n\r\n# Get the vector for a word\r\nword_vector = model.wv['couchbase']\r\n\r\n# Print the vector\r\nprint(word_vector)\r\n\r\nThis Python code would output something like:\r\n\r\n[-0.00053675, 0.000236998, 0.00510486, 0.00900848, ..., 0.000901757, 0.00639282]<\/pre>\n
Sentence and Document Embeddings<\/b>: Moving beyond single words, sentence and document embeddings represent larger pieces of text. These embeddings can capture the context of an entire sentence or document, not just individual words. Models like BERT and Doc2Vec are good examples. They\u2019re used in tasks that require understanding the overall message, sentiment, or topic of texts.<\/span><\/p>\n
Image Embeddings<\/b>: These convert images into vectors, capturing visual features like shapes, colors, and textures. Image embeddings are created using deep learning models (like <\/span>CNNs: Convolutional Neural Networks<\/span><\/a>). They enable tasks such as image recognition, classification, and similarity searches. For example, an image embedding might help a computer recognize whether a given picture is a hot dog or not.<\/span><\/p>\n
Graph Embeddings<\/b>: Graph embeddings are used to represent relationships and structures, such as social networks, org charts, or biological pathways. They turn the nodes and edges of a graph into vectors, capturing how items are connected. This is useful for recommendations, clustering, and detecting communities (clusters) within networks.<\/span><\/p>\n
Audio Embeddings<\/b>: Similar to image embeddings, audio embeddings translate sound into vectors, capturing features like pitch, tone, and rhythm. These are used in voice recognition, music analysis, and sound classification tasks.<\/span><\/p>\n
Video Embeddings<\/b>: Video embeddings capture both the visual and temporal dynamics of videos. They\u2019re used for activities like video search, classification, and understanding scenes or activities within the footage.<\/span><\/p>\n
How to Create Vector Embeddings<\/h2>\n
Generally speaking, there are four steps:<\/span><\/p>\n
    \n
  1. \n
      \n
    1. Choose Your Vector Embedding Model<\/b>: Decide on the type of model based on your needs. Word2Vec, GloVe, and FastText are popular for word embeddings, while BERT and GPT-4 are used for sentence and document embeddings, etc.<\/span><\/li>\n
    2. Prepare Your Data<\/b>: Clean and preprocess your data. For text, this can include tokenization, removing “stopwords,” and possibly lemmatization (reducing words to their base form). For images, this might include resizing, normalizing pixel values, etc.<\/span><\/li>\n
    3. Train or Use Pre-trained Models<\/b>: You can train your model on your dataset or use a pre-trained model. Training from scratch requires a significant amount of data, time, and computational resources. Pre-trained models are a quick way to get started and can be fine-tuned (or augmented) with your specific dataset.<\/span><\/li>\n
    4. Generate Embeddings<\/b>: Once your model is ready, feed your data through it (via SDK, REST, etc.) to generate embeddings. Each item will be transformed into a vector that represents its semantic meaning. Typically, the embeddings are stored in a database, sometimes right alongside the original data.<\/span><\/li>\n<\/ol>\n<\/li>\n<\/ol>\n
      Applications of Vector Embeddings<\/h2>\n
      So, what\u2019s the big deal with vector? What problems can I attack with it? Here are several use cases that are enabled by using vector embeddings to find semantically similar items (i.e., “vector search”):<\/span><\/p>\n
      Natural Language Processing (NLP)<\/h3>\n