How relevance is determined in Hybrid Search
Hybrid Search at Cludo enhances the quality and accuracy of your search results by combining keyword and vector search methods. This approach improves how the Cludo search engine understands what users really mean — even if they don’t use the exact words.
What is Hybrid Search?
Hybrid Search blends two powerful techniques:
- Keyword search (Lucene-based): Matches the exact words in your query.
- Vector search (semantic-based): Understands the meaning and context of your words, not just the literal text.
By combining both, Cludo hybrid search finds results that are both exact and contextually relevant.
Why use Vector Search?
Vector search makes search smarter by analyzing the context of your query.
Example
For example, the word “apple” could mean:
- A fruit
- A tech company
A keyword search might just look for the word “apple,” but vector search looks at the whole sentence to figure out what you really mean.
Let’s compare:
- Query: “How do I update my Apple?”
Interpretation: Likely referring to a phone or computer. - Query: “How do I plant an apple?”
Interpretation: Referring to a fruit.
Even if you don’t say “iPhone,” “MacBook,” or “fruit,” the system can figure it out from context.
How it works
Hybrid Search at Cludo uses a method called Reciprocal Rank Fusion (RRF) to combine the search results of a keyword search and a vector search for a given query. To do so, each result is given a Rank Score based on how high it ranks in each set of search results. The results are then sorted by their Rank Scores, and a final result list is provided back to the end user.
Note about Rank Score: This scoring is based on the position of the result in each list — not the actual relevance score numbers.
In summary, Hybrid Search improves the relevancy of your search results by combining traditional keyword matching with context analyzing power. Using this, Cludo Hybrid Search can deliver balanced, relevant results that understand what you mean, not just what you type.