Granted Patent B2

US 10,754,881 · App. 15/429,483 · Granted Aug 25, 2020

Inventors: Vassilis Plachouras (London, GB); Jochen Lothar Leidner (London, GB); Charese Smiley (Eagan, MN); Hiroko Bretz (Roseville, MN)

Assignee: REFINITIV US ORGANIZATION LLC

G06F16/3329G06Q40/00

Abstract

Claims (34)

1. A computer-implemented method comprising:

identifying a first set of entities corresponding to an indexed data set in response to a user query, wherein the indexed data set includes indexed virtual documents that have been generated using one or more unindexed data sets, wherein the indexed virtual documents are generated and indexed by applying one or more templates from a set of predefined set of templates to the one or more unindexed data sets;

generating a ranked list of query intents using the first set of entities, wherein each query intent of the list of query intents represents a second set of entities associated with the user query, and wherein the generating a ranked list of query intents includes:

computing a score for each query intent of the list of query intents, wherein computing a score for a query intent of the list of query intents is based at least in part on a prevalence of at least one entity of the second set of entities represented by the query intent across the second sets of entities represented by each of the list of query intents; and

ranking each query intent of the list of query intents based on the computed score to generate the ranked list of query intents;

identifying a query plan from a set of predefined query plans;

iterating over the ranked list of query intents to identify a top ranked intent associated to the query plan;

executing the query plan associated with the top ranked intent using the set of entities corresponding to the top ranked intent, the query plan comprising one or more search actions against the indexed data set;

receiving a first set of results in response to one or more search actions;

selecting a language template for text generation based on the first set of results;

generating a description of the first set of results using the language template; and

transmitting the description in response to the user query.

2. The computer-implemented method of claim 1 , wherein the second set of entities associated with the user query is generated by disambiguation of the first set of entities.

3. The computer-implemented method of claim 1 , wherein the query plan further comprises at least one of filtering results according to a specified condition, ranking results according to the values of a given field or aggregating associated values.

4. The computer-implemented method of claim 1 , wherein the description further comprises an overview and context of the first set of results.

5. The computer-implemented method of claim 1 , wherein generating the description of the first set of results further comprises populating the language template with one or more data items from the first set of results.

6. A system comprising:

a non-transient memory; and

a server including a processor configured to:

identify a first set of entities corresponding to an indexed data set in response to a user query, wherein the indexed data set includes indexed virtual documents that have been generated using one or more unindexed data sets, wherein the indexed virtual documents are generated and indexed by applying one or more templates from a set of predefined set of templates to the one or more unindexed data sets;

generate a ranked list of query intents using the first set of entities, wherein each query intent of the list of query intents represents a second set of entities associated with the user query, and wherein the configuration of the processor to generate a ranked list of query intents includes configuration of the processor to:

compute a score for each query intent of the list of query intents, wherein computing a score for a query intent of the list of query intents is based at least in part on a prevalence of at least one entity of the second set of entities represented by the query intent across the second sets of entities represented by each of the list of query intents; and

rank each query intent of the list of query intents based on the computed score to generate the ranked list of query intents;

identify a query plan from a set of predefined query plans;

iterate over the ranked list of query intents to identify a top ranked intent associated to the query plan;

execute the query plan associated with the top ranked intent using the set of entities corresponding to the top ranked intent, the query plan comprising one or more search actions against the indexed data set;

receive a first set of results in response to one or more search actions;

select a language template for text generation based on the first set of results;

generate a description of the first set of results; and

transmit the description in response to the user query.

7. The system of claim 6 , wherein the second set of entities associated with the user query is generated by disambiguation of the first set of entities.

8. The system of claim 6 , wherein the query plan further comprises at least one of filtering results according to a specified condition, ranking results according to the values of a given field or aggregating associated values.

9. The system of claim 6 , wherein the description further comprises an overview and context of the first set of results.

10. The system of claim 6 , wherein generating the description of the first set of results further comprises populating the language template with one or more data items from the first set of results.

Assignments (11)

Continuity (2)