Boosting query in Elasticsearch - Step-by-Step Execution

Concept Flow - Boosting query

Start: Input Query

↓

Define Positive Query

↓

Define Negative Query

↓

Set Boost Factor

↓

Execute Boosting Query

↓

Results with Scores Adjusted

↓

End

The boosting query runs a positive query to find relevant documents, then reduces scores of documents matching a negative query by a boost factor, returning adjusted results.

Execution Sample

Elasticsearch

{
  "query": {
    "boosting": {
      "positive": { "match": { "title": "apple" } },
      "negative": { "match": { "title": "fruit" } },
      "negative_boost": 0.5
    }
  }
}

This query finds documents with 'apple' in the title but lowers scores for those also containing 'fruit' by half.

Execution Table

Step	Action	Query Part	Documents Matched	Score Effect
1	Run positive query	match title: apple	Doc1, Doc2, Doc3	Base scores assigned
2	Run negative query	match title: fruit	Doc2, Doc4	Identify docs to reduce score
3	Apply negative boost	negative_boost: 0.5	Doc2	Score of Doc2 multiplied by 0.5
4	Combine scores	boosting query	Doc1, Doc2, Doc3	Doc1 and Doc3 keep base score; Doc2 score reduced
5	Return results	final output	Doc1, Doc2, Doc3	Sorted by adjusted scores
6	End	-	-	Execution complete

💡 All documents matching positive query processed; negative boost applied to overlapping docs; results returned.

Variable Tracker

Variable	Start	After Step 1	After Step 2	After Step 3	Final
positive_docs	[]	[Doc1, Doc2, Doc3]	[Doc1, Doc2, Doc3]	[Doc1, Doc2, Doc3]	[Doc1, Doc2, Doc3]
negative_docs	[]	[]	[Doc2, Doc4]	[Doc2, Doc4]	[Doc2, Doc4]
scores	{}	{Doc1:1.0, Doc2:1.0, Doc3:1.0}	{Doc1:1.0, Doc2:1.0, Doc3:1.0}	{Doc1:1.0, Doc2:0.5, Doc3:1.0}	{Doc1:1.0, Doc2:0.5, Doc3:1.0}

Key Moments - 3 Insights

Why does Doc2's score get reduced but Doc4 is not in the final results?

Doc2 matches both positive and negative queries, so its score is reduced by the negative boost (see execution_table step 3). Doc4 only matches the negative query, so it is excluded because it doesn't match the positive query (step 1).

What happens if the negative_boost is set to 1?

If negative_boost is 1, scores of documents matching the negative query are multiplied by 1, so their scores stay the same (no reduction). This means the boosting query behaves like a normal positive query without score reduction (refer to execution_table step 3).

Can documents not matching the positive query appear in results?

No, only documents matching the positive query are considered for results. Negative query only reduces scores of these documents if they also match the negative query (see execution_table steps 1 and 4).

Visual Quiz - 3 Questions

Look at the execution table, what is the score of Doc2 after applying the negative boost?

A 1.0

B 0.0

C 0.5

D 2.0

💡 Hint

Check execution_table row 3 where negative boost is applied to Doc2's score.

At which step does the query exclude documents that do not match the positive query?

A Step 1

B Step 4

C Step 2

D Step 5

💡 Hint

Look at execution_table step 1 where positive query matches documents.

If negative_boost was set to 0, what would happen to the scores of documents matching the negative query?

A Scores would be doubled

B Scores would be zeroed out

C Scores would be unchanged

D Documents would be removed

💡 Hint

Negative boost multiplies scores; 0 means scores become zero (see execution_table step 3 logic).

Concept Snapshot

Boosting query syntax:
{
  "query": {
    "boosting": {
      "positive": {...},
      "negative": {...},
      "negative_boost": 0.5
    }
  }
}

Behavior:
- Finds docs matching positive query
- Reduces scores of docs matching negative query by negative_boost
- Returns results sorted by adjusted scores

Key rule: Only docs matching positive query appear; negative query lowers their scores.

Full Transcript

A boosting query in Elasticsearch runs two queries: a positive query to find relevant documents and a negative query to identify documents whose scores should be reduced. The negative_boost factor multiplies the scores of documents matching the negative query, lowering their rank. Documents not matching the positive query are excluded. This process helps prioritize documents matching the positive query but penalizes those also matching the negative query. The execution table shows step-by-step how documents are matched and scores adjusted. The variable tracker follows the sets of documents and their scores through each step. Key moments clarify common confusions about which documents appear and how scores change. The visual quiz tests understanding of score changes and query behavior. The concept snapshot summarizes the syntax and main behavior of the boosting query.