Apparatus and method for matching a query to partitioned document path segments
AssigneeRaining Data Corporation
InventorsJeff Dexter, Allen James Ratcliffe, Eric Soirot
Life of this patent
Estimating the term…
Abstract
A method of answering a query includes deconstructing documents into path segments. Identical path segments from different documents are arranged contiguously to form a set of partitions, where each partition has identical path segments. The structure of a query is then analyzed to find a match with a document. Based upon the analysis, a sub-set of partitions is selected for searching. Content from the query is compared to content within the sub-set of partitions to identify matched content. The matched content has associated structural attributes. The matched content structural attributes are compared to the structure of the query to identify an answer to the query.
Claims35
A method of answering a query, comprising:
- deconstructing documents into path segments;
- arranging identical path segments from different documents contiguously to form a set of partitions, wherein each partition has identical path segments;
- analyzing the structure of a query;
- selecting, based upon said analyzing, a sub-set of partitions for searching;
- comparing content from said query to content within said sub-set of partitions to identify matched content, said matched content having associated structural attributes;
- and
- determining whether said matched content structural attributes correspond to the structure of said query to identify an answer to said query.
A method of answering a query, comprising:
- partitioning documents into path segments;
- arranging identical path segments from different documents contiguously to form a set of partitions, wherein each partition has identical path segments;
- mapping a query to a selected partition of said set of partitions;
- designating content within said query that matches content in said selected partition as matched content;
- and
- supplying an answer to said query when said matched content structurally corresponds to said query.
A computer readable medium, comprising:
- a document partitioner with executable instructions to
- deconstruct documents into path segments, and
- arrange identical path segments from different documents contiguously to form a set of partitions, wherein each partition has identical path segments;
- and
- a query processor with executable instructions to
- analyze the structure of a query,
- select a sub-set of partitions to search,
- match content from said query to content within said sub-set of partitions to identify matched content, said matched content having associated structural attributes, and
- compare said matched content structural attributes to the structure of said query to identify an answer to said query.
Dependencies are inferred from the claim text.