Why Completion suggester in Elasticsearch? - Purpose & Use Cases

The Big Idea

What if your search box could guess what users want instantly, making shopping a breeze?

The Scenario

Imagine you run a large online store and want to help customers find products quickly as they type in the search box.

Without any special tool, you try to guess what they might want by checking every product name manually.

The Problem

Manually scanning thousands or millions of product names for every keystroke is very slow and can freeze your website.

It also leads to mistakes, like missing good suggestions or showing irrelevant ones.

The Solution

The Completion suggester in Elasticsearch quickly finds matching suggestions as users type, using a special index built for fast lookups.

This makes search boxes smart and responsive, showing relevant options instantly.

Before vs After

✗ Before

search all products for prefix 'lap' every time user types

✓ After

{
  "suggest": {
    "product-suggest": {
      "prefix": "lap",
      "completion": {
        "field": "suggest_field"
      }
    }
  }
}

What It Enables

It enables lightning-fast, relevant autocomplete suggestions that improve user experience and boost sales.

Real Life Example

When you start typing 'lap' in a store search box, it instantly suggests 'laptop', 'lap desk', or 'lapel microphone' without delay.

Key Takeaways

Manual search for suggestions is slow and error-prone.

Completion suggester uses a special index for instant prefix matching.

This makes search boxes fast, smart, and user-friendly.

Practice

1. What is the main purpose of the completion suggester in Elasticsearch?

easy

A. To update documents in bulk

B. To perform full-text search on documents

C. To aggregate data based on terms

D. To provide fast autocomplete suggestions as users type

Solution

1. Step 1: Understand the role of completion suggester

   The completion suggester is designed to quickly suggest words or phrases as a user types input.

2. Step 2: Differentiate from other Elasticsearch features

   Full-text search, aggregation, and bulk updates serve different purposes unrelated to autocomplete suggestions.

3. Final Answer:

   To provide fast autocomplete suggestions as users type -> Option D

4. Quick Check:

   Completion suggester = autocomplete suggestions [OK]

Hint: Completion suggester = fast autocomplete feature [OK]

Common Mistakes:

• Confusing completion suggester with full-text search
• Thinking it aggregates data
• Assuming it updates documents

2. Which of the following is the correct way to define a completion field in an Elasticsearch mapping?

easy

A. {
"properties": {
"suggest": { "type": "completion" }
}
}

B. {
"properties": {
"suggest": { "type": "text" }
}
}

C. {
"mappings": {
"suggest": { "type": "completion" }
}
}

D. {
"fields": {
"suggest": { "type": "keyword" }
}
}

Solution

1. Step 1: Identify correct mapping structure

   The completion field must be defined inside properties with type set to completion.

2. Step 2: Compare options

   { "properties": { "suggest": { "type": "completion" } } } correctly uses properties and type: completion. { "properties": { "suggest": { "type": "text" } } } uses text type, which is incorrect. { "mappings": { "suggest": { "type": "completion" } } } wrongly places suggest directly under mappings. { "fields": { "suggest": { "type": "keyword" } } } uses fields and keyword type, which is incorrect.

3. Final Answer:

   { "properties": { "suggest": { "type": "completion" } } } -> Option A

4. Quick Check:

   Completion field in properties with type completion = { "properties": { "suggest": { "type": "completion" } } } [OK]

Hint: Completion field must be under properties with type 'completion' [OK]

Common Mistakes:

• Placing completion field outside properties
• Using text or keyword type instead of completion
• Misnaming mapping keys

3. Given the following suggest query, what will be the expected output if the index contains documents with suggest fields 'apple', 'applet', and 'application'?

{
  "suggest": {
    "fruit-suggest": {
      "prefix": "app",
      "completion": {
        "field": "suggest"
      }
    }
  }
}

medium

A. ["apple", "applet", "application"]

B. ["app", "apple", "applet"]

C. ["application"]

D. []

Solution

1. Step 1: Understand the prefix and completion suggester

   The prefix 'app' matches all suggest entries starting with 'app', which includes 'apple', 'applet', and 'application'.

2. Step 2: Predict the output list

   The completion suggester returns all matching suggestions starting with the prefix, so all three words will be suggested.

3. Final Answer:

   ["apple", "applet", "application"] -> Option A

4. Quick Check:

   Prefix 'app' matches all three suggestions [OK]

Hint: Prefix matches all suggestions starting with it [OK]

Common Mistakes:

• Expecting partial matches inside words
• Thinking prefix must be exact full word
• Assuming empty or partial results

4. Identify the error in this suggest query:

{
  "suggest": {
    "my-suggest": {
      "prefix": "ban",
      "completion": {
        "field": "suggestion"
      }
    }
  }
}

Assuming the mapping defines the completion field as suggest (not suggestion).

medium

A. The prefix value 'ban' is invalid

B. The suggest name 'my-suggest' is not allowed

C. The field name in completion should be 'suggest', not 'suggestion'

D. The suggest query must be inside 'query' key

Solution

1. Step 1: Check field name in completion suggester

   The completion suggester must reference the correct field name defined in the mapping. Here, the mapping uses 'suggest', but the query uses 'suggestion'.

2. Step 2: Validate other parts of the query

   The prefix 'ban' is valid, suggest name 'my-suggest' is allowed, and suggest queries do not require a 'query' key.

3. Final Answer:

   The field name in completion should be 'suggest', not 'suggestion' -> Option C

4. Quick Check:

   Field name mismatch causes error [OK]

Hint: Field name in query must match mapping exactly [OK]

Common Mistakes:

• Using wrong field name in completion query
• Thinking prefix value is invalid
• Expecting suggest query inside 'query' key

5. You want to implement a completion suggester that suggests city names but excludes any suggestions that start with the letter 'x'. Which approach correctly applies this filter in the suggest query?

hard

A. {
"suggest": {
"city-suggest": {
"prefix": "",
"completion": {
"field": "city_suggest",
"skip_duplicates": true
}
}
}
}

B. {
"suggest": {
"city-suggest": {
"prefix": "",
"completion": {
"field": "city_suggest",
"contexts": {
"first_letter": ["a", "b", "c", "d", "e", "f", "g", "h", "i", "j", "k", "l", "m", "n", "o", "p", "q", "r", "s", "t", "u", "v", "w", "y", "z"]
}
}
}
}
}

C. {
"suggest": {
"city-suggest": {
"prefix": "",
"completion": {
"field": "city_suggest",
"contexts": {
"first_letter": ["!x"]
}
}
}
}
}

D. {
"suggest": {
"city-suggest": {
"prefix": "",
"completion": {
"field": "city_suggest",
"filter": {
"term": { "first_letter": "x" }
}
}
}
}
}

Solution

1. Step 1: Understand filtering with contexts in completion suggester

   Completion suggester supports filtering using contexts. To exclude suggestions starting with 'x', you include all allowed first letters except 'x' in the context filter.

2. Step 2: Analyze each option

   { "suggest": { "city-suggest": { "prefix": "", "completion": { "field": "city_suggest", "skip_duplicates": true } } } } has no filtering. { "suggest": { "city-suggest": { "prefix": "", "completion": { "field": "city_suggest", "contexts": { "first_letter": ["!x"] } } } } } tries to exclude 'x' with '!x' which is invalid syntax. { "suggest": { "city-suggest": { "prefix": "", "completion": { "field": "city_suggest", "contexts": { "first_letter": ["a", "b", "c", "d", "e", "f", "g", "h", "i", "j", "k", "l", "m", "n", "o", "p", "q", "r", "s", "t", "u", "v", "w", "y", "z"] } } } } } explicitly lists all letters except 'x' in contexts, correctly filtering out 'x'. { "suggest": { "city-suggest": { "prefix": "", "completion": { "field": "city_suggest", "filter": { "term": { "first_letter": "x" } } } } } } uses a 'filter' key which is not supported in completion suggester.

3. Final Answer:

   { "suggest": { "city-suggest": { "prefix": "", "completion": { "field": "city_suggest", "contexts": { "first_letter": ["a", "b", "c", "d", "e", "f", "g", "h", "i", "j", "k", "l", "m", "n", "o", "p", "q", "r", "s", "t", "u", "v", "w", "y", "z"] } } } } } -> Option B

4. Quick Check:

   Use contexts with allowed letters excluding 'x' [OK]

Hint: Use contexts with allowed letters, exclude 'x' by omission [OK]

Common Mistakes:

• Using invalid '!x' syntax in contexts
• Trying to use 'filter' inside completion suggester
• Not applying context filtering at all