CN-122029506-A - Method, apparatus and computer program product for interpretable artificial intelligence of mental model awareness for intelligent user interfaces

CN122029506ACN 122029506 ACN122029506 ACN 122029506ACN-122029506-A

Abstract

A system and method for determining content to recommend to a user interface is provided. The system may determine a context of a user in the environment. The system may implement a machine learning model that includes training data that is pre-trained or that is trained in real-time based on historical interactions of users with the data or interactions of users with content determined in real-time. The system may analyze one or more items of contextual information associated with the context to determine content related to a user associated with a device capturing the content items in the environment. The system may analyze one or more items of context information or other items of context information to determine context variables in the environments that are determined to be relevant to the system. The system may determine one or more recommendations or one or more actions to present to the user interface using the determined content related to the user and the determined context variables that are determined to be related to the system.

Inventors

Kashyapu Todi
Tanya Reni Junko
Benjamin Lafrenier
Thomas. Langrak
Luta Parimar Desai

Assignees

元平台公司

Dates

Publication Date: 20260512
Application Date: 20241015
Priority Date: 20241009

Claims (13)

1.A method, comprising: Determining one or more contexts of one or more users in one or more environments; Implementing a machine learning model, the machine learning model comprising training data, the training data being pre-trained, or the training data being trained in real-time based on historical interactions of one or more other users with the data or interactions of the one or more other users with content determined in real-time; Analyzing at least one item of context information associated with the one or more contexts to determine content related to a user associated with a device capturing content items in an environment; analyzing the at least one item of the context information or other items of the context information to determine one or more context variables in the one or more environments determined to be relevant to the device by implementing the machine learning model, and A recommendation or action to be presented to a user interface is determined using the determined content related to the user and the determined one or more context variables determined to be related to the device.
2. The method of claim 1, wherein the apparatus comprises a head-mounted device.
3. The method of claim 1 or 2, further comprising: Determining whether the at least one item of the context information and the one or more context variables are relevant or irrelevant to presentation via the user interface.
4. A method according to claim 3, further comprising: In response to determining that a corresponding determination score associated with the at least one item of context information and the one or more context variables equals or exceeds a predetermined threshold, determining that the at least one item of context information and the one or more context variables are related.
5. The method of claim 3 or 4, further comprising: Interpretable data is generated, the interpretable data being associated with the at least one item of the contextual information or with at least a subset of one or more items of the determined contextual variables related to the presentation.
6. The method of claim 5, further comprising: the interpretable data is presented to the user interface to enable the user to view or interact with the interpretable data.
7. The method of any preceding claim, further comprising: It is beneficial to determine that the content related to the user is presented through the user interface for consideration by the user even if the user is unaware of the content related to the user.
8. The method of any preceding claim, further comprising: at least one application associated with the determined at least one recommendation for presentation via the user interface is generated.
9. The method of claim 8, further comprising: at least one task associated with the action and the application for presentation via the user interface is generated.
10. The method of any preceding claim, further comprising: determining that the one or more context variables determined to be relevant to the apparatus include detected information associated with a second user, an environment associated with the second user being the same associated environment as the environment associated with the user, or the second user being associated with another of the one or more environments.
11. An apparatus, comprising: One or more processors, and At least one memory storing instructions that, when executed by the one or more processors, cause the apparatus to perform the method of any preceding claim.
12. A computer readable medium storing instructions that when executed by a processor cause the processor to perform the method of any one of claims 1 to 10.
13. A computer program product comprising instructions which, when the program is executed by a computer, cause the computer to perform the method according to any one of claims 1 to 10.

Description

Method, apparatus and computer program product for interpretable artificial intelligence of mental model awareness for intelligent user interfaces Cross Reference to Related Applications The present application claims priority and benefit from U.S. provisional patent application serial No. 63/590,332 filed on day 2023, month 10, and U.S. non-provisional patent application serial No. 18/911,164 filed on day 2024, month 10, 9. Technical Field The present disclosure relates generally to methods, apparatus, and computer program products for facilitating user interface optimization and for providing an interpretable artificial intelligence context-adaptive user interface. Background Currently, intelligent user interfaces (INTELLIGENT USER INTERFACE, IUI) can provide timely, relevant, and personalized experiences for users of different use cases. While some IUIs are easier to use than rule-based systems, these IUIs may tend to lack transparency. This can be overwhelming for the user, resulting in more frustration, longer task time, and reduced confidence for the user. For example, some existing IUI methods typically ignore awareness and assumptions regarding the behavior of the system by the user. In this regard, these existing systems often lack mechanisms to determine how much information about the system model and decisions should be interpreted by the interface. Furthermore, bombing the user with too much or unnecessary detail may be counterproductive to the user. It may therefore be beneficial to provide an efficient and reliable mechanism that provides enhanced techniques to determine which context variables should be included in one or more interpretations to a user interface. Disclosure of Invention Some example aspects of the present disclosure may facilitate improvements in user interface optimization and interpretable artificial intelligence for adaptive user interfaces. Accordingly, some example aspects of the present disclosure may provide systems, methods, and/or manners that may determine mental perceptions of a user about a context associated with the user and may utilize the determined context to determine relevant and timely interpretation and adaptation recommendations provided/presented through a user interface. In this regard, some examples of the present disclosure may provide context-adaptive user interfaces with mental model aware interpretable artificial intelligence (ARTIFICIAL INTELLIGENCE, AI). Exemplary aspects of the present disclosure may utilize sensitivity analysis to determine the importance of a context variable, thereby facilitating an interpretable AI user interface. Exemplary aspects may determine criteria such as which context variables one or more users may misunderstand or be uncertain about which context variables may be important to the user and, in some examples, whether one or more users may be surprised by one or more adjustments that may be provided by the user interface. Criteria may be provided for analysis by an optimization scheme for the system of the present disclosure to enable the system to determine dense interpretations in a high-dimensional environment, thereby providing these interpretations for an adaptive user interface. According to a first aspect, a method is provided that includes determining one or more contexts of one or more users in one or more environments, implementing a machine learning model that includes training data that is pre-trained or that is trained in real-time based on historical interactions of one or more other users with data or interactions of the one or more other users with content determined in real-time, analyzing at least one item of context information associated with the one or more contexts to determine content related to the user that is associated with a device capturing content items in the environment, analyzing the at least one item of context information or other items of context information to determine one or more context variables in the one or more environments that are determined to be related to the device by implementing the machine learning model, and utilizing the determined content related to the user and the determined one or more context variables determined to be related to the device to determine a recommendation action to be presented to the user interface. The apparatus may comprise a head mounted device. The method may also include determining whether the at least one item of context information and the one or more context variables are related or not related to presentation via the user interface. The method may further include determining that the at least one item of context information and the one or more context variables are related in response to determining that a corresponding determination score associated with the at least one item of context information and the one or more context variables equals or exceeds a predetermined threshold. The method may further include generating interpreta