# Common Uses of Vector Embeddings with text-embedding-ada-002 **Semantic Analysis** https://docs.pinecone.io/page/examples ## [[Retrieval Augmented Generation (RAG)]] ## Search Where results are ranked by relevance to a query string. Search engines traditionally work by searching for overlaps of keywords. By leveraging vector embeddings, [semantic search](https://docs.pinecone.io/docs/semantic-text-search) can go beyond keyword matching and deliver based on the query’s semantic meaning. - **Semantic search**: Embeddings help computers understand the meaning behind what people are looking for, making search results more relevant. - https://cookbook.openai.com/examples/semantic_text_search_using_embeddings - https://docs.pinecone.io/docs/semantic-text-search - **Query expansion**: Embeddings can identify related words and phrases, making searches more comprehensive and accurate. - **Image search:** vector embeddings are perfectly suited to serve as the basis for image retrieval tasks. There are multiple off-the-shelf models, such as [CLIP](https://www.pinecone.io/learn/series/image-search/clip/), ResNet, and more. Different models handle different types of tasks like [image similarity](https://docs.pinecone.io/docs/image-similarity-search), object detection, and many more. - **Audio search:** by converting the audio into a set of activations (an audio spectrogram), we produce vector embeddings that can be used for [audio similarity search](https://docs.pinecone.io/docs/audio-search). ## Clustering Where data is grouped by similarity. Clustering is one way of making sense of a large volume of textual data. Embeddings are useful for this task, as they provide semantically meaningful vector representations of each text. Thus, in an unsupervised way, clustering will uncover hidden groupings in our dataset. - **Topic modeling**: Grouping texts based on their meaning, using embeddings to find similarities. - **Visual analytics**: Making complex data easier to see and understand by showing it in simpler, lower-dimensional spaces. https://cookbook.openai.com/examples/clustering ### Recommendation Systems Where items with related data are recommended. Because shorter distances between embedding vectors represent greater similarity, embeddings can be useful for recommendation. We can create embeddings out of structured data that correlate to different entities such as [products](https://docs.pinecone.io/docs/product-recommendation-engine), articles, etc. In most cases, you’d have to create your own embedding model since it would be specific to your particular application. Sometimes this can be combined with unstructured embedding methods when images or text descriptions are found. - **Collaborative filtering**: Using embeddings to understand user preferences and recommend items based on similarities. - **Content-based filtering**: Identifying similar content by representing items and users with embeddings, allowing for personalized recommendations. ## Anomaly Detection Where outliers with little relatedness are identified. We can create embeddings for anomaly detection using large data sets of labeled sensor information that [identify anomalous occurrences](https://docs.pinecone.io/docs/it-threat-detection). - **Outlier detection**: Finding unusual data points by comparing their embeddings to the rest of the dataset, helping to detect fraud, spam, or other malicious activities. - **Predictive maintenance**: Monitoring equipment behavior using embeddings and detecting anomalies that might signal potential failures or issues. ## Diversity Measurement Where similarity distributions are analyzed. - **Document diversity**: Analyzing text data using embeddings to measure the variety of topics, styles, or viewpoints within a set of documents. - **Population diversity**: Assessing the diversity of a group, like employees, based on various attributes, such as skills or background, by converting these attributes into embeddings. ## Classification Where text strings are classified by their most similar label. - **Sentiment analysis**: Assigning emotions, like positive, negative, or neutral, to text by using embeddings as input features for machine learning algorithms or deep learning models, like neural networks. - **Labeling**: Categorizing text or data points based on their embeddings, which helps in tasks like organizing information, filtering content, or tagging items for easier retrieval.