Planning-based automated fusing of data from multiple heterogeneous sources

摘要

Methods and software for automatedly answering dynamic queries requiring fusing of data from various ones of multiple available heterogeneous data sources. The data sources are characterized and interrelated using a suitable ontology. In one implementation, each dynamic query is analyzed to determine the data items required to answer that query, along with their levels of detail. The data items and levels of detail are used to determine which ones of the multiple data sources are needed to answer the query. The data items, levels of detail, and the ontology of the data sources are then used to automatedly develop a data-fusing plan for fusing data from the needed data sources in a manner that answers the query. The data-fusing plan is executed, and the query is answered.

主权项

1. A machine-readable storage medium containing machine-executable instructions for performing a method of answering a dynamic query having an answer that requires the fusing of data from among multiple heterogeneous data sources, said machine-executable instructions comprising:
a first set of machine-executable instructions for receiving the dynamic query; a second set of machine-executable instructions for automatedly evaluating the dynamic query to determine data items and levels of detail needed to provide the answer; a third set of machine-executable instructions for automatedly developing a data-fusing plan based on an ontology of the multiple heterogeneous data sources as a function of the data items and levels of detail, wherein the data-fusion plan includes one or more steps for creating fused data; a fourth set of machine-executable instructions for automatedly executing the one or more steps to create the fused data based on the data items; and a fifth set of machine-executable instructions for automatedly generating the answer based on the fused data; wherein:
said third set of machine-executable instructions includes machine-executable instructions for implementing a graph theoretic approach to determine which ones of the multiple heterogeneous data sources are needed to generate the answer;said machine-executable instructions for implementing the graph theoretic approach includes machine-executable instructions for constructing a graph of the multiple heterogeneous data sources;said machine-executable instructions for implementing the graph theoretic further includes machine-executable instructions for traversing the graph so as to determine a set needed data sources needed data to create the fused data; andsaid machine-executable instructions for traversing the graph includes machine-executable instructions for:
starting with a needed data source:next selecting an adjacent connected needed data source, if available; andif none of the adjacent data sources is a needed data source, next selecting a more highly connected adjacent data source over a less highly connected adjacent data source.