In addition to the metadata that Graphlit indexes into the knowledge graph, all the text from your content is automatically added into a searchable index.
With the contents query, you can use the search field to provide text to search by.
Graphlit uses OpenAI embedding models to create vector embeddings of the text extracted from your content. The vector embeddings are added into a built-in vector database for knowledge retrieval.
Your search text also has a vector embedding created, which gets matched against the stored vector embeddings to find the most similar query results.
Vector-based Similarity Search
Here is an example of finding similar content to the search text "Unstructured Data":
Query:
query QueryContents($filter: ContentFilter!) {
contents(filter: $filter) {
results {
id
name
creationDate
state
owner {
id
}
originalDate
finishedDate
workflowDuration
uri
text
type
fileType
mimeType
fileName
fileSize
masterUri
mezzanineUri
transcriptUri
}
}
}
