Documentation
¶
Overview ¶
Package graph provides graph-related utilities for RAG systems. It includes components for extracting entities and relationships from text, as well as searching knowledge graphs for relevant information.
Package graph provides graph-related utilities for RAG systems. It includes components for extracting entities and relationships from text, as well as searching knowledge graphs for relevant information.
Index ¶
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
This section is empty.
Types ¶
type GlobalSearcher ¶
type GlobalSearcher struct {
// contains filtered or unexported fields
}
GlobalSearcher performs Map-Reduce over Community Summaries for macro-level questions. It helps answer broad questions by synthesizing information from multiple communities in the knowledge graph.
func NewGlobalSearcher ¶
func NewGlobalSearcher(store abstraction.GraphStore, llm core.Client) *GlobalSearcher
NewGlobalSearcher creates a new global searcher.
Parameters: - store: The graph store to use for searching - llm: The LLM client to use for synthesizing results
Returns: - A new GlobalSearcher instance
func (*GlobalSearcher) Search ¶
func (s *GlobalSearcher) Search(ctx context.Context, query string, communityLevel int) (string, error)
Search performs Map-Reduce over Community Summaries for macro-level questions.
Parameters: - ctx: The context for the operation - query: The query to answer - communityLevel: The level of community summaries to use
Returns: - A synthesized answer based on community summaries - An error if searching fails
type GraphExtractor ¶
type GraphExtractor struct {
// contains filtered or unexported fields
}
GraphExtractor uses an LLM to extract Entities (Nodes) and Relationships (Edges) from text chunks. It helps build knowledge graphs from unstructured text for better retrieval and reasoning.
func DefaultGraphExtractor ¶
func DefaultGraphExtractor(llm core.Client) *GraphExtractor
DefaultGraphExtractor 创建默认的图提取器 使用 gochat 的标准 LLM 客户端进行实体和关系提取
func NewGraphExtractor ¶
func NewGraphExtractor(llm core.Client) *GraphExtractor
NewGraphExtractor creates a new graph extractor.
Parameters: - llm: The LLM client to use for extraction
Returns: - A new GraphExtractor instance
func (*GraphExtractor) Extract ¶
func (e *GraphExtractor) Extract(ctx context.Context, chunk *entity.Chunk) ([]abstraction.Node, []abstraction.Edge, error)
Extract extracts entities and relationships from a text chunk.
Parameters: - ctx: The context for the operation - chunk: The text chunk to extract from
Returns: - A slice of nodes (entities) - A slice of edges (relationships) - An error if extraction fails
type LocalSearcher ¶
type LocalSearcher struct {
// contains filtered or unexported fields
}
LocalSearcher performs N-Hop traversal from specific entities to gather relational context. It helps retrieve relevant information by traversing the knowledge graph from given entities.
func NewLocalSearcher ¶
func NewLocalSearcher(store abstraction.GraphStore) *LocalSearcher
NewLocalSearcher creates a new local searcher.
Parameters: - store: The graph store to use for searching
Returns: - A new LocalSearcher instance
func (*LocalSearcher) Search ¶
func (s *LocalSearcher) Search(ctx context.Context, entities []string, maxHops int, topK int) (string, error)
Search performs N-Hop traversal from specific entities to gather relational context.
Parameters: - ctx: The context for the operation - entities: The entities to start the search from - maxHops: The maximum number of hops to traverse - topK: The maximum number of results to return
Returns: - A string representation of the search results - An error if searching fails
Source Files