Boost
The boost mapping parameter is used to increase or decrease the relevance score of a field during search queries. It allows you to apply more or less weight to specific fields when calculating the overall relevance score of a document.
The boost parameter is applied as a multiplier to the score of a field. For example, if a field has a boost value of 2, then the score contribution of that field is doubled. Conversely, a boost value of 0.5 would halve the score contribution of that field.
When using the boost parameter, it is recommended that you start with small values (1.5 or 2) and test the effect on your search results. Overly high boost values can skew the relevance scores and lead to unexpected or undesirable search results.
The boost parameter only applies to term-level queries. It does not apply to prefix, range, or fuzzy term-level queries.
Index-time and query-time boosting
While you can set boost values in field mappings (index-time boosting), this is not recommended. Instead, use query-time boosting, which offers several advantages:
-
Flexibility: Query-time boosting allows you to adjust boost values without reindexing documents.
-
Precision: Index-time boosts are stored as part of the
norms, which uses only 1 byte. This can reduce the resolution of field length normalization. -
Dynamic control: Query-time boosting gives you the ability to experiment with different boost values for different use cases.
Example
Use the boost parameter to give more weight to certain fields. For instance, boosting the title field more than the description field can improve results if the title is a stronger indicator of relevance.
In this example, the title field has a boost of 2, so it contributes twice as much to the relevance score than the description field (which has a default boost of 1).
Index-time boosting (not recommended)
Create an index with boosted fields (for demonstration purposes only):
PUT /article_index
{
"mappings": {
"properties": {
"title": {
"type": "text",
"boost": 2
},
"description": {
"type": "text"
}
}
}
}
Add some sample documents to the index:
PUT /article_index/_doc/1
{
"title": "Introduction to Machine Learning",
"description": "This article covers basic algorithms and their applications in data science."
}
PUT /article_index/_doc/2
{
"title": "Data Science Fundamentals",
"description": "Learn about machine learning algorithms and statistical methods for analyzing data."
}
Search across both fields using index-time boosting:
POST /article_index/_search
{
"query": {
"multi_match": {
"query": "machine learning algorithms",
"fields": ["title", "description"]
}
}
}
Document 1 is scored higher because “machine learning” appears in the boosted title field. Document 2 is scored lower because “machine learning” appears in the unboosted description field. Both documents contain “algorithms”, which contributes to their scores:
{
"took": 415,
"timed_out": false,
"_shards": {
"total": 1,
"successful": 1,
"skipped": 0,
"failed": 0
},
"hits": {
"total": {
"value": 2,
"relation": "eq"
},
"max_score": 1.1906823,
"hits": [
{
"_index": "article_index",
"_id": "1",
"_score": 1.1906823,
"_source": {
"title": "Introduction to Machine Learning",
"description": "This article covers basic algorithms and their applications in data science."
}
},
{
"_index": "article_index",
"_id": "2",
"_score": 0.7130072,
"_source": {
"title": "Data Science Fundamentals",
"description": "Learn about machine learning algorithms and statistical methods for analyzing data."
}
}
]
}
}
Query-time boosting (recommended)
Instead of index-time boosting, use query-time boosting for better control and flexibility. Query-time boosting doesn’t require any special field mappings to be configured.
Add some sample documents to the index:
PUT /article_index_2/_doc/1
{
"title": "Introduction to Machine Learning",
"description": "This article covers basic algorithms and their applications in data science."
}
PUT /article_index_2/_doc/2
{
"title": "Data Science Fundamentals",
"description": "Learn about machine learning algorithms and statistical methods for analyzing data."
}
First, search the title field without boosting:
POST /article_index_2/_search
{
"query": {
"match": {
"title": {
"query": "machine learning algorithms"
}
}
}
}
The matching document has a score of 0.59:
{
"took": 13,
"timed_out": false,
"_shards": {
"total": 1,
"successful": 1,
"skipped": 0,
"failed": 0
},
"hits": {
"total": {
"value": 1,
"relation": "eq"
},
"max_score": 0.59534115,
"hits": [
{
"_index": "article_index_2",
"_id": "1",
"_score": 0.59534115,
"_source": {
"title": "Introduction to Machine Learning",
"description": "This article covers basic algorithms and their applications in data science."
}
}
]
}
}
Next, search the same field with boosting:
POST /article_index_2/_search
{
"query": {
"match": {
"title": {
"query": "machine learning algorithms",
"boost": 2
}
}
}
}
The document score is doubled:
{
"took": 16,
"timed_out": false,
"_shards": {
"total": 1,
"successful": 1,
"skipped": 0,
"failed": 0
},
"hits": {
"total": {
"value": 1,
"relation": "eq"
},
"max_score": 1.1906823,
"hits": [
{
"_index": "article_index_2",
"_id": "1",
"_score": 1.1906823,
"_source": {
"title": "Introduction to Machine Learning",
"description": "This article covers basic algorithms and their applications in data science."
}
}
]
}
}
To get a baseline for searching both the title and description fields, first search without boosting:
POST /article_index_2/_search
{
"query": {
"multi_match": {
"query": "machine learning algorithms",
"fields": ["title", "description"]
}
}
}
Document 2 is scored higher than Document 1:
{
"took": 10,
"timed_out": false,
"_shards": {
"total": 1,
"successful": 1,
"skipped": 0,
"failed": 0
},
"hits": {
"total": {
"value": 2,
"relation": "eq"
},
"max_score": 0.7130072,
"hits": [
{
"_index": "article_index_2",
"_id": "2",
"_score": 0.7130072,
"_source": {
"title": "Data Science Fundamentals",
"description": "Learn about machine learning algorithms and statistical methods for analyzing data."
}
},
{
"_index": "article_index_2",
"_id": "1",
"_score": 0.59534115,
"_source": {
"title": "Introduction to Machine Learning",
"description": "This article covers basic algorithms and their applications in data science."
}
}
]
}
}
To compare index-time boosting with query-time boosting, search multiple fields with query-time boosting:
POST /article_index_2/_search
{
"query": {
"multi_match": {
"query": "machine learning algorithms",
"fields": ["title^2", "description"]
}
}
}
This query produces the same response as the index-time boosting query, with Document 1 now scoring higher than Document 2:
{
"took": 5,
"timed_out": false,
"_shards": {
"total": 1,
"successful": 1,
"skipped": 0,
"failed": 0
},
"hits": {
"total": {
"value": 2,
"relation": "eq"
},
"max_score": 1.1906823,
"hits": [
{
"_index": "article_index_2",
"_id": "1",
"_score": 1.1906823,
"_source": {
"title": "Introduction to Machine Learning",
"description": "This article covers basic algorithms and their applications in data science."
}
},
{
"_index": "article_index_2",
"_id": "2",
"_score": 0.7130072,
"_source": {
"title": "Data Science Fundamentals",
"description": "Learn about machine learning algorithms and statistical methods for analyzing data."
}
}
]
}
}