Search-as-you-type field in Elasticsearch - Step-by-Step Execution

Concept Flow - Search-as-you-type field

User types a letter

↓

Send partial input to Elasticsearch

↓

Elasticsearch matches partial input

↓

Return matching suggestions

↓

Display suggestions to user

↓

User continues typing or selects suggestion

↓

User types a letter

The search-as-you-type field sends partial user input to Elasticsearch, which returns matching suggestions instantly as the user types.

Execution Sample

Elasticsearch

PUT my_index
{
  "mappings": {
    "properties": {
      "title": {
        "type": "search_as_you_type"
      }
    }
  }
}

This code creates an index with a search-as-you-type field called 'title' that supports fast partial matching.

Execution Table

Step	User Input	Query Sent to Elasticsearch	Matching Documents Found	Suggestions Displayed
1	"a"	{"match": {"title": "a"}}	['apple', 'apricot']	['apple', 'apricot']
2	"ap"	{"match": {"title": "ap"}}	['apple', 'apricot']	['apple', 'apricot']
3	"app"	{"match": {"title": "app"}}	['apple', 'application']	['apple', 'application']
4	"appl"	{"match": {"title": "appl"}}	['apple', 'application']	['apple', 'application']
5	"apple"	{"match": {"title": "apple"}}	['apple']	['apple']
6	"apple " (user stops typing)	No new query sent	No change	Suggestions remain ['apple']

💡 User stops typing or selects a suggestion, ending the search-as-you-type interaction.

Variable Tracker

Variable	Start	After 1	After 2	After 3	After 4	After 5	Final
User Input	""	"a"	"ap"	"app"	"appl"	"apple"	"apple"
Query Sent	None	{"match": {"title": "a"}}	{"match": {"title": "ap"}}	{"match": {"title": "app"}}	{"match": {"title": "appl"}}	{"match": {"title": "apple"}}	No new query
Suggestions	[]	['apple', 'apricot']	['apple', 'apricot']	['apple', 'application']	['apple', 'application']	['apple']	['apple']

Key Moments - 3 Insights

Why does Elasticsearch send a new query on every letter typed?

Because the search-as-you-type field is designed to update suggestions instantly as the user types, so each new letter triggers a new query (see execution_table steps 1-5).

What happens if the user stops typing?

No new query is sent and the suggestions stay the same until the user types again or selects a suggestion (see execution_table step 6).

Why do suggestions change as more letters are typed?

Because the query becomes more specific, filtering matches to better fit the typed input (see execution_table rows 1 to 5).

Visual Quiz - 3 Questions

Look at the execution_table at step 3, what suggestions are displayed?

A ['apple', 'application']

B ['apple', 'apricot']

C ['apricot', 'application']

D ['apple']

💡 Hint

Check the 'Suggestions Displayed' column at step 3 in the execution_table.

At which step does the query sent to Elasticsearch become the most specific?

A Step 1

B Step 5

C Step 3

D Step 6

💡 Hint

Look at the 'Query Sent to Elasticsearch' column and find the longest user input.

If the user types 'appl' instead of 'app', how would the suggestions change at step 4?

A Suggestions would include 'apricot' only

B Suggestions would be empty

C Suggestions would include 'apple' and 'application'

D Suggestions would include 'apple' only

💡 Hint

Refer to the suggestions shown at step 4 in the execution_table.

Concept Snapshot

Search-as-you-type field in Elasticsearch:
- Use type 'search_as_you_type' in mapping.
- Sends partial input queries as user types.
- Returns matching suggestions instantly.
- Updates suggestions with each new letter.
- Stops querying when user stops typing or selects.

Full Transcript

This visual execution shows how a search-as-you-type field works in Elasticsearch. When the user types each letter, a query is sent to Elasticsearch with the current partial input. Elasticsearch returns matching documents that contain words starting with that input. Suggestions update instantly as the user types more letters, becoming more specific. If the user stops typing, no new queries are sent and suggestions remain. This helps users find matches quickly by showing results as they type.

Practice

1. What is the main purpose of a search_as_you_type field in Elasticsearch?

easy

A. To provide instant search results as the user types each character

B. To store large binary files efficiently

C. To perform complex aggregations on numeric data

D. To index documents only after full text is entered

Solution

1. Step 1: Understand the purpose of search_as_you_type

   The search_as_you_type field is designed to help users get search results instantly as they type, improving user experience.

2. Step 2: Compare with other options

   Options B, C, and D describe unrelated Elasticsearch features like binary storage, aggregations, or delayed indexing, which are not related to search_as_you_type.

3. Final Answer:

   To provide instant search results as the user types each character -> Option A

4. Quick Check:

   search_as_you_type = instant search while typing [OK]

Hint: Remember: search_as_you_type means live search suggestions [OK]

Common Mistakes:

• Confusing search_as_you_type with binary or numeric fields
• Thinking it delays indexing until full input
• Assuming it is for aggregations

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

easy

A. "title": { "type": "text" }

B. "title": { "type": "completion" }

C. "title": { "type": "keyword" }

D. "title": { "type": "search_as_you_type" }

Solution

1. Step 1: Identify the correct field type for search-as-you-type

   The search_as_you_type field type must be explicitly set as "type": "search_as_you_type" in the mapping.

2. Step 2: Eliminate other field types

   "title": { "type": "text" } uses text, which is for full text search but not optimized for search-as-you-type. "title": { "type": "keyword" } uses keyword, which is for exact matches. "title": { "type": "completion" } uses completion, which is for autocomplete but different from search_as_you_type.

3. Final Answer:

   "title": { "type": "search_as_you_type" } -> Option D

4. Quick Check:

   Mapping type for search-as-you-type = search_as_you_type [OK]

Hint: Use type "search_as_you_type" exactly in mapping [OK]

Common Mistakes:

• Using "text" instead of "search_as_you_type"
• Confusing with "completion" type
• Omitting the type field

3. Given this mapping snippet:

{
  "mappings": {
    "properties": {
      "name": { "type": "search_as_you_type" }
    }
  }
}

Which query will correctly search for documents where the name field starts with "mic"?

medium

A. { "match": { "name": "mic" } }

B. { "multi_match": { "query": "mic", "type": "bool_prefix", "fields": ["name", "name._2gram", "name._3gram"] } }

C. { "term": { "name": "mic" } }

D. { "prefix": { "name": "mic" } }

Solution

1. Step 1: Understand how to query search_as_you_type fields

   Search-as-you-type fields require a multi_match query with type: bool_prefix and fields including the base field and its n-gram subfields.

2. Step 2: Analyze each query option

   { "match": { "name": "mic" } } uses simple match, which does not support prefix matching well. { "term": { "name": "mic" } } uses term query, which requires exact match. { "prefix": { "name": "mic" } } uses prefix query, which is not optimized for search_as_you_type. { "multi_match": { "query": "mic", "type": "bool_prefix", "fields": ["name", "name._2gram", "name._3gram"] } } correctly uses multi_match with bool_prefix and the required fields.

3. Final Answer:

   { "multi_match": { "query": "mic", "type": "bool_prefix", "fields": ["name", "name._2gram", "name._3gram"] } } -> Option B

4. Quick Check:

   Use multi_match + bool_prefix + ngram fields [OK]

Hint: Use multi_match with bool_prefix and ngram fields [OK]

Common Mistakes:

• Using term or prefix queries alone
• Not including ngram subfields in fields list
• Using simple match query for prefix search

4. You wrote this query to search a search_as_you_type field but get no results:

{ "multi_match": { "query": "app", "type": "bool_prefix", "fields": ["title"] } }

What is the likely problem?

medium

A. The field 'title' must be of type 'keyword' for this query

B. The 'bool_prefix' type is invalid for multi_match queries

C. The query should include ngram subfields like 'title._2gram' and 'title._3gram'

D. The query string 'app' is too short to match anything

Solution

1. Step 1: Check query fields for search_as_you_type

   Search-as-you-type fields generate subfields with ngrams like '_2gram' and '_3gram' that must be included in the query fields for prefix matching.

2. Step 2: Identify missing ngram subfields in the query

   The query only searches 'title' field, missing 'title._2gram' and 'title._3gram', so it won't match prefixes properly.

3. Final Answer:

   The query should include ngram subfields like 'title._2gram' and 'title._3gram' -> Option C

4. Quick Check:

   Include ngram subfields in fields list for prefix search [OK]

Hint: Always add _2gram and _3gram subfields in multi_match fields [OK]

Common Mistakes:

• Omitting ngram subfields in fields list
• Thinking bool_prefix is invalid
• Assuming keyword type is needed
• Blaming query length

5. You want to implement a search-as-you-type feature for a product catalog with a name field. Which of the following mapping and query combinations will correctly support prefix matching as the user types?

hard

A. Mapping: { "name": { "type": "search_as_you_type" } }Query: { "multi_match": { "query": "lap", "type": "bool_prefix", "fields": ["name", "name._2gram", "name._3gram"] } }

B. Mapping: { "name": { "type": "text" } }Query: { "match": { "name": "lap" } }

C. Mapping: { "name": { "type": "keyword" } }Query: { "prefix": { "name": "lap" } }

D. Mapping: { "name": { "type": "completion" } }Query: { "term": { "name": "lap" } }

Solution

1. Step 1: Choose the correct mapping for search-as-you-type

   The search_as_you_type type is designed for prefix matching and instant search as the user types.

2. Step 2: Use the correct query for prefix matching

   The multi_match query with bool_prefix type and fields including the base field and its ngram subfields is required for effective prefix matching.

3. Step 3: Evaluate other options

   Mapping: { "name": { "type": "text" } }
   Query: { "match": { "name": "lap" } } uses text with simple match, which does not support prefix search well. Mapping: { "name": { "type": "keyword" } }
   Query: { "prefix": { "name": "lap" } } uses keyword with prefix query, which is less flexible. Mapping: { "name": { "type": "completion" } }
   Query: { "term": { "name": "lap" } } uses completion with term query, which is incorrect usage.

4. Final Answer:

   Mapping: { "name": { "type": "search_as_you_type" } } and Query: { "multi_match": { "query": "lap", "type": "bool_prefix", "fields": ["name", "name._2gram", "name._3gram"] } } -> Option A

5. Quick Check:

   search_as_you_type + multi_match bool_prefix + ngram fields = correct [OK]

Hint: Combine search_as_you_type mapping with multi_match bool_prefix query [OK]

Common Mistakes:

• Using text type with match query for prefix search
• Using keyword type with prefix query only
• Confusing completion type with search_as_you_type
• Using term query for prefix matching