Closed-loop natural language query pre-processor and response synthesizer architecture

Patent metadata

Field	Value
Publication number	US-2018144064-A1
Application number	US-201615357574-A
Country	US
Kind code	A1
Filing date	Nov 21, 2016
Priority date	Nov 21, 2016
Publication date	May 24, 2018
Grant date	—

Abstract

A closed-loop natural language query pre-processor and response synthesizer architecture accepts natural language queries and dynamically synthesizes query results. The query results may be in the form of data stories. The architecture identifies, selects, and composes candidate response elements into a coherent and meaningful query result. The architecture also implements an adaptable delivery mechanism that is responsive to connection bandwidth, query source preferences, query source characteristics, and other factors. Feedback from multiple sources adapts the architecture for handling subsequent queries.

First claim

What is claimed is:

1 . A method comprising:
in a query processing hardware system:
obtaining a natural language input query from a query source over a communication interface;
with a query pre-processing architecture:
submitting the natural language input query to a question handler;
determining query components from the natural language input query;
executing searches based on the query components against pre-defined data stores to determine candidate response elements; and
communicating the candidate response elements to a response synthesizer;

with the response synthesizer:
synthesizing a query response to the natural language input query based on the candidate response elements.

2 . The method of claim 1 , where:
determining query components comprises determining an entity explicitly included in the natural language input query, an entity implied but not explicitly included in the natural language input query, or both types of entities.

3 . The method of claim 1 , where:
determining query components comprises determining a requesting entity who has posed the natural language input query; and responsive to determining the requesting entity, obtaining requester metadata that characterizes the requesting entity.

4 . The method of claim 3 , where:
obtaining requester metadata comprises querying an enterprise database for any combination of one or more of:
a requesting entity role;
a requesting entity query history with the query processing hardware system; and
a requesting entity enterprise activity.

5 . The method of claim 1 , where:
the query response package comprises multiple output types arranged according to the element sequence.

6 . The method of claim 5 , where:
the multiple output types include any combination of one or more of:
text output;
hypertext output;
voice output;
interactive visualization output;
video output; and
simulated physical presence output.

7 . The method of claim 1 , further comprising:
determining a recommend action in addition to the candidate response elements; and adding the recommended action into the query response for delivery to the query source.

8 . The method of claim 1 , further comprising:
ranking the candidate response elements according to relevance; and filtering the candidate response elements with a configurable relevance cut-off to obtain the selected element subset.

9 . The method of claim 8 , where:
ranking comprises:
determining measures of the candidate response elements on a multi-dimensional analysis framework, the multi-dimensional analysis framework comprising a ‘relevance’ dimension and an ‘expected engagement’ dimension; and
evaluating a function of the measures on the ‘relevance’ dimension and the ‘expected engagement’ dimension to obtain an overall rank for each of the candidate response elements.

10 . The method of claim 1 , further comprising:
establishing a query session for the query source with a session controller in response to obtaining the natural language input query;

11 . The method of claim 1 , where:
the pre-defined data stores comprise an internal product data store with respect to a pre-defined enterprise that hosts the query processing hardware system; and where: executing searches against the pre-defined data stores comprises: executing the searches against the internal product data store to discover candidate response elements regarding a particular enterprise product.

12 . The method of claim 1 , where:
the pre-defined data stores comprise an internal user data store with respect to a pre-defined enterprise that hosts the query processing hardware system; and where: executing searches against the pre-defined data stores comprises: executing the searches against the internal user data store to discover candidate response elements regarding a particular enterprise individual.

12 . The method of claim 1 , where:
the pre-defined data stores comprise an external data store with respect to a pre-defined enterprise that hosts the query processing hardware system; and where: executing searches against the pre-defined data stores comprises: executing the searches against the external data store to determine the candidate response elements.

13 . The method of claim 1 , further comprising:
determining a selected element subset of the candidate response elements; determining an element sequence for the candidate response elements in the selected element subset; determining a query response package structure; and ordering the candidate response elements in the selected element subset according to the element sequence and the query response package structure to synthesize the query response to the natural language input query.

14 . The method of claim 1 , where:
synthesizing includes: applying pre-determined weightings to the candidate response elements.

15 . A natural language query processing system comprising:
communication interface circuitry configured to connect to a query source and obtain a natural language input query; query pre-processing circuitry configured to:
establish a query session for the query source with a session controller in response to obtaining the natural language input query;
submit the natural language input query to a question handler configured to determine query components from the natural language input;
execute searches based on the query components against pre-defined data stores to determine candidate response elements; and
communicate the candidate response elements to response synthesis circuitry; and
response synthesis circuitry configured to:
determine a selected element subset of the candidate response elements;
determine an element sequence for the candidate response elements in the selected element subset;
determine a query response package structure;
order the candidate response elements in the selected element subset according to the element sequence and the query response package structure to synthesize a query response to the natural language input query; and
deliver the query response through the session controller to the query source via the communication interface.

16 . The system of claim 15 , where:
the query components comprise:
an entity explicitly included in the natural language input query;
an entity implied but not explicitly included in the natural language input query; or
both types of entities.

17 . The system of claim 15 , where:
query pre-processing circuitry is further configured to:
determine a requesting entity that has posed the natural language input query; and
responsive to determining the requesting entity, obtain requester metadata that characterizes the requesting entity.

18 . The system of claim 17 , where:
query pre-processing circuitry is further configured to: obtain the requester metadata by querying an enterprise database for any combination of one or more of:
a requesting entity role;
a requesting entity query history with the natural language query processing system; and
a requesting entity enterprise activity.

19 . The system of claim 15 , where:
the query response package comprises multiple output types arranged according to the element sequence.

20 . The system of claim 19 , where:
the multiple output types include any combination of one or more of:
text output;
hypertext output;
voice output;
interactive visualization output;
video output; and
simulated p

Assignees

• Accenture Global Solutions Ltd

Inventors

• Krasadakis Georgios

Classifications

• G10L15/00

  Speech recognition (G10L17/00 takes precedence) · CPC title

• G06F16/9038

  Presentation of query results · CPC title

• G06F40/30

  Semantic analysis · CPC title

• G10L15/22

  Procedures used during a speech recognition process, e.g. man-machine dialogue · CPC title

• G10L15/1822

  Parsing for meaning understanding · CPC title