US 11,531,822 B1Granted

Training models and using the trained models to indicate staleness of content items

AssigneeAmazon Technologies, Inc.

InventorsMingwen Dong, Sheng-Min Shih, Tapodipta Ghosh, Mihir Anil Joshi, Stuart Myles, Kun Cao

Life of this patent

Estimating the term…

Loading patent…

Abstract

A model training service of a provider network may receive content items as training data. For example, the content items may be documents with certain portions labeled as stale. The model training service may train one or more different types of models using those content items (e.g., natural language inference model, paraphrasing detection model, named entity recognition model). The model training service may then provide the model(s) to a content staleness check (CSC) service. The CSC service may receive, from a client, a request that indicates one or more content items to be checked for staleness. The CSC service may process the content items by the model(s) to generate one or more indications of staleness of the content item. The CSC service may generate a response based on the indications of staleness. For example, the CSC service may generate a document with certain portions highlighted as stale content.

Claims20

Dependent claims

A system, comprising:

one or more processors;
and
one or more memories, wherein the one or more memories have stored thereon program instructions for a model training service, which when executed by one or more processors, cause the one or more processors to:
obtain a plurality of content items for training one or more models for use by a content staleness check service (CSC service);
train the one or more models, using the plurality of content items, to determine staleness of content items, wherein staleness of a content item indicates whether the content item is out-of-date;
and
provide the one or more trained models for deployment to the CSC service;
and
one or more other processors;
and
one or more other memories, wherein the one or more other memories have stored thereon program instructions for the CSC service, which when executed by one or more other processors, cause the one or more other processors to:
receive, from a client, a request indicating one or more other content items to be checked for staleness, wherein for a given other content item, the CSC service is configured to:
process the other content item by the one or more trained models to generate an indication of staleness of the other content item;
and
generate a response based at least on the indications of staleness of the one or more other content items generated by the one or more trained models;
and
send the response to the client.

14 more elements

A method, comprising:

performing, by one or more computing devices:
receiving one or more content items;
processing the one or more content items by one or more trained models to generate one or more indications of staleness of at least one of the content items, wherein the one or more models are trained based on a plurality of other content items to determine staleness of content items, and wherein staleness of a content item indicates whether the content item is out-of-date;
generating a result based on the one or more indications of staleness of the at least one content item generated by the one or more trained models;
and
sending the result to a destination.

3 more elements

One or more non-transitory computer-accessible storage media storing program instructions that when executed on or across one or more processors of a provider network cause the one or more processors to:

obtain a plurality of content items for training one or more models;
train the one or more models, using the plurality of content items, to determine staleness of content items, wherein staleness of a content item indicates whether the content item is out-of-date;
and
provide the one or more trained models for deployment to a destination, wherein the one or more trained models are configured to process other content items to generate indications of staleness of the other content items.

1 more element

Dependencies are inferred from the claim text.