US 12,182,526 B2Granted

Interactive graphical interfaces for efficient localization of natural language generation responses, resulting in natural and grammatical target language output

AssigneeGoogle Inc.

InventorsKatherine Vadella, Joshua Andrews, Max Copperman, Gabrielle Gayles, Shanjian Li, Jieyu Lu, Luchuan Xu

Life of this patent

Estimating the term…

Loading patent…

Abstract

Implementations relate to effectively localizing system responses, that include dynamic information, to target language(s), such that the system responses are grammatical and/or natural in the target language(s). Some of those implementations relate to various techniques for resource efficient generation of templates for a target language. Some versions of those implementations relate to resource efficient generation of target language natural language generation (NLG) templates and, more particularly, to techniques that enable a human user to generate a target language NLG template more efficiently and/or with greater accuracy. The more efficient target language NLG template generation enables less utilization of various client device resources and/or can mitigate the risk of flawed NLG templates being provided for live use in one or more systems.

Claims20

Dependent claims

A method implemented by one or more processors, the method comprising:

identifying, from a database, a source language natural language generation (NLG) template for a source language,
wherein the source language NLG template comprises natural language source static text that is in the source language and comprises a natural language source language primitive that defines an argument;
determining, from a corpus of candidate entities and based on defined attributes of a target language, a subset of candidate entities for the argument;
generating a graphical user interface that includes:
a source language NLG portion including:
the natural language source language primitive, and
a natural language source language example including the natural language source static text and a candidate entity, of the subset of candidate entities, in the source language, and in lieu of the natural language source language primitive, wherein the candidate entity semantically corresponds with the natural language source static text;
and
an interactive target language NLG template portion including:
a target language translation of the natural language source static text, and
a modifiable element corresponding to natural language selectable candidate target language primitives that are unique and simultaneously displayed upon initial rendering of the interactive target language NLG template portion and that are each a candidate for
utilization, in a target NLG template, in lieu of the modifiable element;
causing the graphical user interface to be rendered at a client device;
receiving user input directed to a given target language primitive, of the unique simultaneously displayed natural language selectable candidate target language primitives, the user input being via one or more user interface inputs, at the client device, that are directed to the interactive target language NLG template portion;
and
in response to receiving the user input directed to the given target language primitive:
generating a plurality of unique output examples that are in the target language and that each include a natural language target language translation of the source static text and a corresponding dynamic text for the given target language primitive, the corresponding dynamic texts each being generated based on a respective entity of the subset of candidate entities, and each being generated based on any additional constraints imposed by the given target language primitive;
and
causing the plurality of unique output examples to be simultaneously rendered at the client device in the graphical user interface.

16 more elements

A method implemented by one or more processors, the method comprising:

identifying, from a database, a source language natural language generation (NLG) template for a source language,
wherein the source language NLG template comprises natural language source static text that is in the source language and comprises a natural language source language primitive that defines a taxonomic classification;
determining, based on the taxonomic classification and/or the source language static text, a corresponding score for each of a plurality of natural language selectable candidate target language primitives;
determining a presentation order, of the natural language selectable candidate target language primitives, based on the score;
generating a graphical user interface that includes:
a source language NLG portion including:
the natural language source language primitive, and
a source language example including the natural language source static text and a candidate entity in lieu of the natural language source language primitive;
an interactive target language NLG template portion, wherein the interactive target language NLG template portion includes:
a target language translation of the natural language source static text,
a modifiable element corresponding to the natural language selectable candidate target language primitives, in the presentation order, that are unique and simultaneously displayed upon initial rendering of the NLG template portion, in lieu of the source language primitive;
causing the graphical user interface to be rendered at a client device;
receiving a user selection of a given target language primitive, of the unique and simultaneously displayed natural language selectable candidate target language primitives, the user selection being via one or more user interface inputs, at the client device, that are directed to the interactive target language NLG template portion;
and
in response to receiving the user selection of the given target language primitive:
generating a plurality of unique output examples that are in the target language and that each include the target language translation of the natural language source static text and a corresponding dynamic text for the given target language primitive;
and
causing the plurality of unique output examples to be simultaneously rendered at the client device in the graphical user interface.

15 more elements

A method implemented by one or more processors, the method comprising:

identifying, from a database, a source language natural language generation (NLG) template for a source language,
wherein the source language NLG template comprises natural language source static text that is in the source language;
generating a graphical user interface that includes:
a plurality of natural language source output examples that are in the source language, that include the natural language source static text and an entity corresponding to a source language primitive, and that are each generated based on the source language NLG template;
an interactive target language NLG template portion, wherein the interactive target language NLG template portion includes:
a modifiable element corresponding to selectable natural language candidate target language primitives, for the target language, that are unique and simultaneously displayed upon initial rendering of the NLG template portion and that are each a candidate for utilization, in a target NLG template that corresponds to the source language target NLG template, in lieu of the source language primitive;
causing the graphical user interface to be rendered at a client device;
receiving user input directed to a given target language primitive, of the unique simultaneously displayed natural language candidate target language primitives, the user input being via one or more user interface inputs, at the client device, that are directed to the interactive target language NLG template portion;
and
in response to receiving the user input directed to selection of the given natural language target language primitive:
generating a plurality of unique target output examples that are in the target language and that each include a target language translation of the natural language source static text and a corresponding natural language dynamic text for the given natural language target language primitive, the corresponding natural language dynamic texts each being generated based on a respective entity selected based on defined attributes of the target language;
and
causing the plurality of unique target output examples to be simultaneously rendered at the client device, in the graphical user interface, and along with the source output examples.

10 more elements

Dependencies are inferred from the claim text.

Facts

Application: 17/318,704Filed 12 May 2021
Published: US 2022/0366154 A117 Nov 2022
Granted: 31 Dec 2024
Term ends: Not estimatedThe source records don’t support an estimate.

Classification7 codes

G06F 40/58
Processing or translation of natural language: use of machine translation, e.g. for multi-lingual retrieval, for server-side translation for…
G06F 40/186
Editing, e.g. inserting or deleting: templates
G06F 40/47
Data-driven translation: machine-assisted translation, e.g. using translation memory
G06N 5/02
Computing arrangements using knowledge-based models: knowledge representation; Symbolic representation

Family

US 2022/0366154 A1Pre-grant publication · published 17 Nov 2022
US 12,182,526 B2This patent

From the US filing history. Foreign family members aren’t listed.

Timeline

Citations

Open elsewhere

Building an agent or pulling many patents? Use the API or MCP server instead of scraping pages. Get an API key