A database to store index and embeddings of text, image and videos to find out similar content.

Embeddings are numeric representation of text, image and videos.

Semantic_search: Not a streightforward keyword search its a search by understanding the intent of user query

Sample of embeddings

Vector context keyRevenue of AppleCalories in AppleNutrition in Orange
related_to_phone100
is_location000
has_stock100
revenue8000
is_fruit011
calories09580

If we see vector value of calories in apple and nutrition in orange. they have a matching values of is_fruit. the similarities are calculated based on the similarities of values in of the vector of text context.

There are different technics of generating vector embeddings. like Word2Vec, BERT, GPT, ELMo, Fast text etc