README
¶
Apache Answer Plugin: pgvector Vector Search
This plugin enables semantic/vector search in Apache Answer using PostgreSQL with the pgvector extension.
Prerequisites
- PostgreSQL 12+ with the pgvector extension installed
- An OpenAI-compatible embedding API (e.g., OpenAI, Azure OpenAI, or any compatible provider)
Installation
Build Apache Answer with this plugin:
./answer build --with github.com/apache/answer-plugins/vector-search-pgvector
Configuration
After enabling the plugin in the Admin UI (Admin > Plugins > Vector Search), configure the following fields:
| Field | Description | Example |
|---|---|---|
| DSN | PostgreSQL connection string | postgres://user:pass@localhost:5432/answer?sslmode=disable |
| Embedding API Host | OpenAI-compatible API base URL | https://api.openai.com |
| Embedding API Key | API key for the embedding service | sk-... |
| Embedding Model | Model name for generating embeddings | text-embedding-3-small |
| Embedding Level | question embeds question + all answers + comments together; answer embeds each answer separately |
question |
| Similarity Threshold | Minimum cosine similarity score (0-1). Default 0 means no filtering |
0.5 |
How It Works
- Embedding dimensions are auto-detected from the configured model. No manual dimension configuration is needed.
- On first configuration, the plugin creates a table
answer_vector_embeddingswith avectorcolumn matching the detected dimensions. - If the embedding model changes and produces different dimensions, the table is automatically dropped and recreated.
- Uses cosine similarity (
<=>operator) for vector search. - A full sync of all questions/answers is triggered when the plugin starts.
Database Setup
Ensure the pgvector extension is installed in your PostgreSQL database:
CREATE EXTENSION IF NOT EXISTS vector;
License
Documentation
¶
Index ¶
- Variables
- type VectorSearchConfig
- type VectorSearchEngine
- func (e *VectorSearchEngine) ConfigFields() []plugin.ConfigField
- func (e *VectorSearchEngine) ConfigReceiver(config []byte) error
- func (e *VectorSearchEngine) DeleteContent(ctx context.Context, objectID string) error
- func (e *VectorSearchEngine) Description() plugin.VectorSearchDesc
- func (e *VectorSearchEngine) Info() plugin.Info
- func (e *VectorSearchEngine) RegisterSyncer(ctx context.Context, syncer plugin.VectorSearchSyncer)
- func (e *VectorSearchEngine) SearchSimilar(ctx context.Context, query string, topK int) ([]plugin.VectorSearchResult, error)
- func (e *VectorSearchEngine) UpdateContent(ctx context.Context, content *plugin.VectorSearchContent) error
Constants ¶
This section is empty.
Variables ¶
View Source
var Info embed.FS
Functions ¶
This section is empty.
Types ¶
type VectorSearchConfig ¶
type VectorSearchConfig struct {
DSN string `json:"dsn"`
APIHost string `json:"api_host"`
APIKey string `json:"api_key"`
EmbeddingModel string `json:"embedding_model"`
EmbeddingLevel string `json:"embedding_level"`
SimilarityThreshold float64 `json:"similarity_threshold"`
}
VectorSearchConfig holds all plugin configuration.
type VectorSearchEngine ¶
type VectorSearchEngine struct {
Config *VectorSearchConfig
// contains filtered or unexported fields
}
VectorSearchEngine implements plugin.VectorSearch using PostgreSQL + pgvector.
func (*VectorSearchEngine) ConfigFields ¶
func (e *VectorSearchEngine) ConfigFields() []plugin.ConfigField
ConfigFields returns the plugin configuration form fields.
func (*VectorSearchEngine) ConfigReceiver ¶
func (e *VectorSearchEngine) ConfigReceiver(config []byte) error
ConfigReceiver applies configuration from the admin UI.
func (*VectorSearchEngine) DeleteContent ¶
func (e *VectorSearchEngine) DeleteContent(ctx context.Context, objectID string) error
DeleteContent removes a document by object ID.
func (*VectorSearchEngine) Description ¶
func (e *VectorSearchEngine) Description() plugin.VectorSearchDesc
Description returns metadata about this vector search engine.
func (*VectorSearchEngine) Info ¶
func (e *VectorSearchEngine) Info() plugin.Info
func (*VectorSearchEngine) RegisterSyncer ¶
func (e *VectorSearchEngine) RegisterSyncer(ctx context.Context, syncer plugin.VectorSearchSyncer)
RegisterSyncer stores the syncer and triggers a full sync.
func (*VectorSearchEngine) SearchSimilar ¶
func (e *VectorSearchEngine) SearchSimilar(ctx context.Context, query string, topK int) ([]plugin.VectorSearchResult, error)
SearchSimilar performs a cosine similarity search using pgvector.
func (*VectorSearchEngine) UpdateContent ¶
func (e *VectorSearchEngine) UpdateContent(ctx context.Context, content *plugin.VectorSearchContent) error
UpdateContent upserts a single document into the pgvector table.
Source Files
Directories
Click to show internal directories.
Click to hide internal directories.