BaseContextBuilder
BaseContextBuilder
Bases: ABC
Abstract base class for building context using graph data and a search engine.
Attributes:
-
graph–The graph object.
-
single_batch–Whether to process data in a single batch.
-
search_engine–The search engine for retrieving top-k objects.
-
token_encoder–Token encoder for text tokenization.
__init__(graph, single_batch=True, search_engine=None, token_encoder=None)
Initialize the BaseContextBuilder.
Parameters:
-
graph(Graph) –The graph object.
-
single_batch(bool, default:True) –Whether to process data in a single batch.
-
search_engine(Optional[BaseSearchEngine], default:None) –The search engine for similarity searches.
-
token_encoder(Optional[Encoding], default:None) –Token encoder for text tokenization. Defaults to "cl100k_base".
build_context(*args, **kwargs)
abstractmethod
async
Abstract method to build context.
Returns:
-
str | List[str]–The generated context as a string or list of strings.
batch_and_convert_to_text(graph_data, section_name, single_batch=False, max_tokens=12000)
Convert graph data to a formatted string or list of strings in batches based on token count.
Parameters:
-
graph_data(DataFrame) –The graph data to convert.
-
section_name(str) –The section name for the header.
-
single_batch(bool, default:False) –Whether to process data in a single batch. Defaults to False.
-
max_tokens(int, default:12000) –Maximum number of tokens per batch. Defaults to 12000.
Returns:
-
str | List[str]–The formatted graph data as a string or list of strings.
retrieve_top_k_objects(query, k=10, oversample_scaler=2, **kwargs)
async
Retrieve the top-k objects most similar to the query.
Parameters:
-
query(str) –The query string.
-
k(int, default:10) –The number of top results to retrieve. Defaults to 10.
-
**kwargs(Dict[str, Any], default:{}) –Additional parameters for the search engine.
Returns:
-
List[str]–A list of the top-k results.
Raises:
-
ValueError–If
kis less than or equal to 0 or if the search engine is not initialized.