Semantic Search

Semantic search allows you to find documents based on meaning and intent, rather than word matches like full-text does.
Instead of searching for literal occurrences of the words you type, the system analyzes the semantic similarity between the query and the document content. 
To perform this kind of search, you have to enter a text expression.

How Semantic Search Works

1. Document Understanding
   Documents are analyzed and converted into numerical representations (embeddings) that capture their meaning.
   
   
2. Query Interpretation
   Text entered into the search field is converted into an embedding using the same method.
   
   
3. Similarity Matching
   The system compares the query embedding with document embeddings and returns the documents whose meaning is most similar to the query.

Results are ranked based on semantic similarity rather than keyword frequency.

A set of settings may be used to guide the search.

Language

Specifies the language used to interpret both the query and the document content.
Selecting the correct language improves semantic accuracy.

Embedding Scheme

Defines the embedding model or strategy used to calculate semantic similarity.

Folder

Limits semantic search to a specific folder. Leave it blank to search all accessible documents. Enable subfolder search to include all nested folders.

Threshold

Controls how close the semantic match must be for a document to appear in the results.

Lower values lead to broader results, more loosely related documents
Higher values lead to stricter results, closer semantic matches
A value around 50 provides a balanced trade-off between recall and precision.

Search in Current Hits

If enabled, the semantic search is applied only to the documents already returned by a previous search, allowing incremental refinement.

Retrieve Aliases

Includes documents that are aliases of matching documents in the result set.