Package com.redis.vl.extensions.cache

Class SemanticCache

java.lang.Object

com.redis.vl.extensions.cache.BaseCache

com.redis.vl.extensions.cache.SemanticCache

public class SemanticCache
extends BaseCache

Semantic cache for LLM responses using vector similarity. Port of redis-vl-python/redisvl/extensions/cache/llm/semantic.py

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static class	SemanticCache.Builder	Builder for SemanticCache.

Field Summary

Fields inherited from class com.redis.vl.extensions.cache.BaseCache

name, prefix, redisClient, ttl

Method Summary

Modifier and Type	Method	Description
Optional<CacheHit>	check(String prompt)	Check for a semantically similar prompt in the cache.
Optional<CacheHit>	check(String prompt, Filter filter)	Check for a semantically similar prompt with filtering.
List<Optional<CacheHit>>	checkBatch(List<String> prompts)	Batch check multiple prompts.
List<CacheHit>	checkTopK(String prompt, int k)	Get top-k similar results.
void	clear(Filter filter)	Clear cache with optional filter.
float	getDistanceThreshold()	Get the current distance threshold.
long	getHitCount()	Get the hit count.
float	getHitRate()	Get the cache hit rate.
long	getMissCount()	Get the miss count.
void	resetStatistics()	Reset statistics counters.
void	setDistanceThreshold(float threshold)	Set the distance threshold for similarity matching.
void	store(String prompt, String response)	Store a prompt-response pair in the cache.
void	store(String prompt, String response, Map<String,Object> metadata)	Store a prompt-response pair with metadata.
void	storeBatch(List<PromptResponsePair> pairs)	Batch store multiple prompt-response pairs.
void	storeWithTTL(String prompt, String response, Map<String,Object> metadata, int ttl)	Store with custom TTL.
void	update(String prompt, String newResponse)	Update an existing cache entry.

Methods inherited from class com.redis.vl.extensions.cache.BaseCache

clear, disconnect, expire, finalize, getName, getPrefix, getTtl, isConnected, makeKey, setTtl, setWithTtl, setWithTtl, size

Methods inherited from class java.lang.Object

clone, equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Details

store

public void store(String prompt,
 String response)

Store a prompt-response pair in the cache.

Parameters:

prompt - The prompt text

response - The response text

store

public void store(String prompt,
 String response,
 Map<String,Object> metadata)

Store a prompt-response pair with metadata.

Parameters:

prompt - The prompt text

response - The response text

metadata - Additional metadata

storeWithTTL

public void storeWithTTL(String prompt,
 String response,
 Map<String,Object> metadata,
 int ttl)

Store with custom TTL.

Parameters:

prompt - The prompt text

response - The response text

metadata - Additional metadata

ttl - Time-to-live in seconds

check

public Optional<CacheHit> check(String prompt)

Check for a semantically similar prompt in the cache.

Parameters:

prompt - The prompt to check

Returns:

Optional containing the cache hit if found

check

public Optional<CacheHit> check(String prompt,
 Filter filter)

Check for a semantically similar prompt with filtering.

Parameters:

prompt - The prompt to check

filter - Optional filter expression

Returns:

Optional containing the cache hit if found

checkTopK

public List<CacheHit> checkTopK(String prompt,
 int k)

Get top-k similar results.

Parameters:

prompt - The prompt to check

k - Number of results to return

Returns:

List of cache hits sorted by distance

update

public void update(String prompt,
 String newResponse)

Update an existing cache entry.

Parameters:

prompt - The prompt to update

newResponse - The new response

storeBatch

public void storeBatch(List<PromptResponsePair> pairs)

Batch store multiple prompt-response pairs.

Parameters:

pairs - List of prompt-response pairs

checkBatch

public List<Optional<CacheHit>> checkBatch(List<String> prompts)

Batch check multiple prompts.

Parameters:

prompts - List of prompts to check

Returns:

List of optional cache hits

clear

public void clear(Filter filter)

Clear cache with optional filter.

Parameters:

filter - Optional filter expression

getDistanceThreshold

public float getDistanceThreshold()

Get the current distance threshold.

Returns:

The distance threshold

setDistanceThreshold

public void setDistanceThreshold(float threshold)

Set the distance threshold for similarity matching.

Parameters:

threshold - The new threshold (0 = exact match, 1 = any match)

getHitCount

public long getHitCount()

Get the hit count.

Returns:

Number of cache hits

getMissCount

public long getMissCount()

Get the miss count.

Returns:

Number of cache misses

getHitRate

public float getHitRate()

Get the cache hit rate.

Returns:

Hit rate as a percentage (0.0 to 1.0)

resetStatistics

public void resetStatistics()

Reset statistics counters.