Blog

Interval queries: why they are true positional queries, and how to transition from Span

Explains how Interval queries are true positional queries and how to transition to them from Span queries.

Search RelevanceHow To
MS
By: Mayya Sharipova

On

Span queries have long been a tool for ordered and proximity search. These are especially useful for specific domains, such as legal or patent search. But the relatively new Interval queries actually fit this job much better. Unlike Span queries, Interval queries are true positional queries that score documents only based on positional proximity (expanded upon below).

Starting from Elasticsearch v8.16, we have brought Interval queries into parity with Span queries. Specifically:

  • Interval queries now support "range" and "regexp" rules.
  • Interval rules based on multiple terms similar to Span queries can expand up to indices.query.bool.max_clause_count terms instead of previous 128 value.

Our future plan is to deprecate Span queries in favor of Intervals queries, which cover the same functional capability but do so in a more user-friendly way.

Advantages of Interval queries over Span queries

Interval queries rank documents based on the order and proximity of matching terms. Some advantages of Interval queries:

  • True positional queries
  • Grounded in academic research, based on the minimal interval semantics paper with proven algorithms that scale linearly with the number of positions
  • Simpler syntax
  • Slightly faster (no need of score calculations based on corpus statistics)
  • Ability to use scripts for specialized use cases

Interval queries are true positional queries and only consider positional information while scoring documents (scores are inversely proportional to interval's length). This is unlike Span queries that also consider standard metrics like TF-IDF. Below is an example that illustrates how interval queries can do better ranking.

PUT docs
{
  "mappings": {
    "properties": {
      "content": {
        "type": "text"
      }
    }
  }
}


PUT docs/_doc/1
{
  "content" : "She sells beautiful seashells by the seashore, their smooth shapes shining in the sun, catching the light with every curve. The girl’s bright smile is just as inviting, drawing people in as they stop to admire the shells, each one a little piece of the ocean she loves. Her gentle voice, like the sound of the waves, adds to the peaceful charm of the moment."
}

PUT docs/_doc/2
{
  "content" : "She plays; her father sells seashells. "
}

We want to find documents where the term "she" is near the term "sells". The desired ranking would return the 1st document followed by the 2nd document, as these terms occur closer to each other in the 1st document than in the second document.

But if we run a Span query, we will get a different ranking: [doc2, doc1], because Span queries in addition to proximity calculations also incorporate corpus stats such as TF and IDF metrics that will distort ranking purely by proximity.

GET docs/_search?explain=true
{
  "query": {
    "span_near": {
      "clauses": [
        {
          "span_term": {
            "content": "she"
          }
        },
        {
          "span_term": {
            "content": "sells"
          }
        }
      ],
      "slop": 10,
      "in_order": true
    }
  }
}

In contrast, Interval queries calculate scores based on proximity and don't consider corpus stats and length of documents. We will get the desired ranking: [doc1, doc2].

GET docs/_search?explain=true
{
  "query": {
    "intervals": {
      "content": {
        "match": {
          "query": "she sells",
          "max_gaps": 10,
          "ordered" : true
        }
      }
    }
  }
}

This makes Interval queries an ideal choice for true proximity queries.

Interval queries allow to extract the proximity score as a signal for the overall relevance score. They are optimised to be mixed with other relevance signals like BM25, for instance:

GET docs/_search
{
    "query": {
        "bool": {
            "must": {
                "match": {
                    "content": {
                        "query": "she sells",
                        "boost": "{{bm25_boost}}"
                    }
                }
            },
            "should": {
                "intervals": {
                    "content": {
                        "match": {
                            "query": "she sells",
                            "max_gaps": 10
                        },
                        "boost": "{{proximity_boost}}"
                    }
                }
            }
        }
    }
}

Note that this could also be applied to rescoring: we can make the first pass with BM25 alone and then add a rescorer with BM25 + Intervals combination.

Note that if we need to model Span queries behaviour in matching and scoring by BM25 and proximity, we can do it by combining interval queries with BM25 queries as must clauses in a boolean query with appropriate boosts set.

Transition guide

Below we show ways to transition from the following Span queries to the equivalent Interval queries:

  • span_containing
  • span_field_masking
  • span_first
  • span_multi
  • span_near
  • span_not
  • span_or
  • span_term
  • span_within
PUT parks
{
  "mappings": {
    "properties": {
      "park": {
        "type": "text"
      },
      "park_rules": {
        "type": "text"
      }
    }
  }
}

PUT parks/_doc/1
{
  "park" : "Sunny Meadows Park",
  "park_rules" : "Children are encouraged to enjoy our playground equipment, including slides, swings, and climbing structures. Feeding the ducks and fish in the pond is allowed, but only with approved feed available at the park office. Children are not permitted to climb trees or enter the park's fountains and water features. Please do not bring glass containers, sharp objects, or personal sports equipment into the park."
}

PUT parks/_doc/2
{
  "park" : "Greenwood Forest Park",
  "park_rules" : "Children are welcome to explore our nature trails, participate in organized activities, and use the designated picnic areas. Picking flowers, disturbing wildlife, or leaving the designated trails is not allowed. Children must be accompanied by an adult when using the park's grills and fire pits. Please refrain from bringing pets, bicycles, or scooters into the park."
}


PUT parks/_doc/3
{
  "park" : "Happy Haven Playground",
  "park_rules" : "Children can enjoy our sandbox, jungle gym, and seesaws, as well as participate in organized games and activities. Running, shouting, or playing rough games near the playground equipment is not permitted. Children must be supervised by an adult at all times and should use the equipment according to their age and size. Please do not bring food, drinks, or chewing gum into the playground area."
}

PUT parks/_doc/4
{
  "park" : "Lakeside Recreation Park",
  "park_rules" : "Children can enjoy fishing at the lake with an adult, using the sports fields for organized games, and playing in the designated play areas. Swimming, wading, or boating in the lake is strictly prohibited. Children must wear appropriate safety gear when using the sports fields and play equipment. Please do not bring alcohol, tobacco products, or illegal substances into the park."
}

PUT parks/_doc/5
{
  "park" : "Adventure Land Park",
  "park_rules" : "Children are encouraged to use our zip lines, ropes courses, and climbing walls under adult supervision and with proper safety equipment. Running, pushing, or engaging in horseplay near the adventure equipment is not allowed. Children must follow all height, weight, and age restrictions for each activity. Please do not bring personal items, such as cell phones or cameras, onto the adventure equipment."
}

SPAN NEAR

GET parks/_search
{
  "query": {
    "span_near": {
      "clauses": [
        {
          "span_term": {
            "park_rules": "prohibited"
          }
        },
        {
          "span_term": {
            "park_rules": "swimming"
          }
        }
      ],
      "slop": 10,
      "in_order": false
    }
  },
  "highlight": {
    "fields": {
      "park_rules": {}
    }
  }
}

GET parks/_search
{
  "query": {
    "intervals": {
      "park_rules": {
        "match": {
          "query": "swimming prohibited",
          "max_gaps": 10,
          "ordered" : false  
        }
      }
    }
  },
  "highlight": {
    "fields": {
      "park_rules": {}
    }
  }
}

SPAN FIRST

GET parks/_search
{
  "query": {
    "span_first": {
      "match": {
        "span_term": { "park_rules": "sandbox" }
      },
      "end": 5
    }
  },
  "highlight": {
    "fields": {
      "park_rules": {}
    }
  }
}


GET parks/_search
{
  "query": {
    "intervals" : {
      "park_rules" : {
        "match" : {
          "query" : "sandbox",
          "filter" : {
            "script" : {
              "source" : "interval.end < 5"
            }
          }
        }
      }
    }
  },
  "highlight": {
    "fields": {
      "park_rules": {}
    }
  }
}

SPAN OR

GET parks/_search
{
  "query": {
    "span_or" : {
      "clauses" : [
        { "span_term" : { "park_rules" : "prohibited" } },
        { "span_near": {"clauses": [{"span_term": {"park_rules": "not"}}, {"span_term": {"park_rules": "allowed"}}], "in_order": true}},
        { "span_near": {"clauses": [{"span_term": {"park_rules": "not"}}, {"span_term": {"park_rules": "permitted"}}], "in_order": true}}
      ]
    }
  },
  "highlight": {
    "fields": {
      "park_rules": {}
    }
  }
}

GET parks/_search
{
  "query": {
    "intervals" : {
      "park_rules" : {
        "any_of" : {
          "intervals" : [
            { "match" : { "query" : "prohibited"} },
            { "match" : { "query" : "not allowed", "ordered" : true } },
            { "match" : { "query" : "not permitted", "ordered" : true } }
           ]
        }
      }
    }
  },
  "highlight": {
    "fields": {
      "park_rules": {}
    }
  }
}

SPAN CONTAINING

GET parks/_search
{
  "query": {
    "span_containing": {
      "little": {
        "span_term": {
          "park_rules": "sports"
        }
      },
      "big": {
        "span_near": {
          "clauses": [
            {
              "span_term": {
                "park_rules": "children"
              }
            },
            {
              "span_term": {
                "park_rules": "park"
              }
            }
          ],
          "slop": 50,
          "in_order": false
        }
      }
    }
  },
  "highlight": {
    "fields": {
      "park_rules": {}
    }
  }
}

GET parks/_search
{
  "query": {
    "intervals": {
      "park_rules": {
        "match": {
          "query": "children park",
          "max_gaps": 50,
          "filter" : {
            "containing" : {
              "match" : {
                "query" : "sports"
              }
            }
          }
        }
      }
    }
  },
  "highlight": {
    "fields": {
      "park_rules": {}
    }
  }
}

SPAN WITHIN

GET parks/_search
{
  "query": {
    "span_within": {
      "little": {
        "span_term": {
          "park_rules": "sports"
        }
      },
      "big": {
        "span_near": {
          "clauses": [
            {
              "span_term": {
                "park_rules": "children"
              }
            },
            {
              "span_term": {
                "park_rules": "park"
              }
            }
          ],
          "slop": 50,
          "in_order": false
        }
      }
    }
  },
  "highlight": {
    "fields": {
      "park_rules": {
      }
    }
  }
}

GET parks/_search
{
  "query": {
    "intervals": {
      "park_rules": {
        "match": {
          "query": "sports",
          "filter" : {
            "contained_by" : {
              "match" : {
                "query" : "children park",
                "max_gaps": 50
              }
            }
          }
        }
      }
    }
  },
  "highlight": {
    "fields": {
      "park_rules": {
        "number_of_fragments": 0
      }
    }
  }
}

SPAN NOT

GET parks/_search
{
  "query": {
    "span_not": {
      "include": {
        "span_term": { "park_rules": "allowed" }
      },
      "exclude": {
        "span_near": {
          "clauses": [
            { "span_term": { "park_rules": "not" } },
            { "span_term": { "park_rules": "allowed" } }
          ],
          "slop": 0,
          "in_order": true
        }
      }
    }
  },
  "highlight": {
    "fields": {
      "park_rules": {}
    }
  }
}

GET parks/_search
{
  "query": {
    "intervals": {
      "park_rules": {
        "match": {
          "query": "allowed",
          "filter": {
            "not_contained_by": {
              "match": {
                "query": "not allowed",
                "max_gaps": 0,
                "ordered" : true
              }
            }
          }
        }
      }
    }
  },
  "highlight": {
    "fields": {
      "park_rules": {}
    }
  }
}

SPAN_MULTI

wildcard

GET parks/_search
{
    "query": {
        "span_multi": {
            "match": {
                "wildcard": {
                    "park_rules": {"value": "sand*" }
                }
            }
        }
    }
}

GET parks/_search
{
    "query": {
        "intervals": {
            "park_rules": {
                "wildcard": {
                    "pattern": "sand*"
                }
            }
        }
    }
}

fuzzy

GET parks/_search
{
    "query": {
        "span_multi": {
            "match": {
                "fuzzy": {
                    "park_rules": {"value": "sandbo" }
                }
            }
        }
    }
}

GET parks/_search
{
    "query": {
        "intervals": {
            "park_rules": {
                "fuzzy": {
                    "term": "sandbo"
                }
            }
        }
    }
}

prefix

GET parks/_search
{
    "query": {
        "span_multi": {
            "match": {
                "prefix": {
                    "park_rules": {"value": "sandbo" }
                }
            }
        }
    }
}

GET parks/_search
{
    "query": {
        "intervals": {
            "park_rules": {
                "prefix": {
                    "prefix": "sandbo"
                }
            }
        }
    }
}

regexp

GET parks/_search
{
    "query": {
        "span_multi": {
            "match": {
                "regexp": {
                    "park_rules": {"value": "sand.*" }
                }
            }
        }
    }
}

GET parks/_search
{
    "query": {
        "intervals": {
            "park_rules": {
                "regexp": {
                    "pattern": "sand.*"
                }
            }
        }
    }
}

range

GET parks/_search
{
    "query": {
        "span_multi": {
            "match": {
                "range": {
                    "park": {
                        "gte" : "a",
                        "lte": "h"
                    }
                }
            }
        }
    }
}

GET parks/_search
{
    "query": {
        "intervals": {
            "park": {
                "range": {
                    "gte" : "a",
                    "lte" : "h"
                }
            }
        }
    }
}

span_field_masking

use use_field of Intervals

GET parks/_search
{
  "query": {
    "span_near": {
      "clauses": [
        {
          "span_term": {
            "park_rules": "nature"
          }
        },
        {
          "span_field_masking": {
            "query": {
              "span_term": {
                "park_rules.stemmed": "trail"
              }
            },
            "field": "park_rules" 
          }
        }
      ],
      "slop": 5
    }
  }
}


GET parks/_search
{
  "query": {
    "intervals" : {
      "park_rules" : {
        "all_of" : {
          "ordered" : true,
          "max_gaps" : 5, 
          "intervals" : [
            {
              "match" : {
                "query" : "nature"
              }
            },
            {
              "match" : {
                "query" : "trail",
                "use_field" : "park_rules.stemmed"
              }
            }
          ]
        }
      }
    }
  }
}

Conclusion

Interval queries is a powerful tool to do true positional search. Try them with expanded functionalities from 8.16 release.

Ready to try this out on your own? Start a free trial.

Want to get Elastic certified? Find out when the next Elasticsearch Engineer training is running!

Related content

Personalized search with LTR
Search Relevance

August 30, 2024

Personalized search with LTR

Learn how to train ranking models that improve search relevance for individual users.

MJ
By: Max Jakob
Looking back: A timeline of vector search innovations
Vector DatabaseSearch Relevance

August 27, 2024

Looking back: A timeline of vector search innovations

Looking back at Elastic's vector search innovations in Elasticsearch and Lucene

KDBT
By: Kathleen DeRusso and Benjamin Trent
Elasticsearch query rules are now generally available
Search Relevance

August 7, 2024

Elasticsearch query rules are now generally available

Introducing the general availability of query rules

KD
By: Kathleen DeRusso
Introducing Learning To Rank (LTR) in Elasticsearch
Search RelevanceHow To

July 15, 2024

Introducing Learning To Rank (LTR) in Elasticsearch

Discover how Learning To Rank (LTR) can help you to improve your search ranking and how to implement it in Elasticsearch.

AF
By: Aurélien Foucret
Semantic reranking in Elasticsearch with retrievers
Vector DatabaseSearch Relevance+1

May 28, 2024

Semantic reranking in Elasticsearch with retrievers

Explore strategies for using semantic reranking to boost the relevance of top search results, including semantic reranking with retrievers.

ADNC
By: Adam Demjen and Nick Chow

Ready to build state of the art search experiences?

Sufficiently advanced search isn’t achieved with the efforts of one. Elasticsearch is powered by data scientists, ML ops, engineers, and many more who are just as passionate about search as your are. Let’s connect and work together to build the magical search experience that will get you the results you want.

Try it yourself