Method and electronic device for recommending content

The method and device address the challenge of uniform content recommendations in multi-user environments by creating a group knowledge graph and calculating connectivity scores to personalize content for groups based on individual contributions and contextual influences, improving recommendation accuracy.

WO2026240265A1PCT designated stage Publication Date: 2026-11-19SAMSUNG ELECTRONICS CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
PCT/KR2025/017066
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2025-05-12
Filing Date
2025-10-24
Publication Date
2026-11-19

AI Technical Summary

Technical Problem

Existing content recommendation systems fail to differentiate between multiple users' individual interests in shared environments, often assigning equal weight to all users and not accounting for users' unexpressed interests.

Method used

A method and electronic device that generate content recommendations for a group of users by constructing a group knowledge graph, calculating connectivity scores, and generating a group profile to reflect collective content preferences, taking into account individual contributions and contextual influences.

Benefits of technology

The solution effectively personalizes content recommendations for groups by accurately reflecting individual user contributions and preferences, enhancing relevance and user satisfaction.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure KR2025017066_19112026_PF_FP_ABST
    Figure KR2025017066_19112026_PF_FP_ABST
Patent Text Reader

Abstract

An Embodiment of the present disclosure may provide a method (600) for generating content recommendations for a group of users. The method may include obtaining a group knowledge graph for a group of users based on a knowledge graph of each user in the group. The method may include obtaining, for each user in the group, a connectivity score vector based on the group knowledge graph. The method may include obtaining a group connectivity score vector based on the connectivity score vectors of each user of the group. The method may include obtaining, based on the group connectivity score vector and the group knowledge graph, a group profile representing collective content preference of the group. The method may include recommending content to the group based on the group profile.
Need to check novelty before this filing date? Find Prior Art

Description

METHOD AND ELECTRONIC DEVICE FOR RECOMMENDING CONTENT

[0001] The present disclosure generally relates to field of content recommendation system. More particularly, the present disclosure relates to a method and an electronic device for recommending content to multi-users in multi-device environments.

[0002] The content recommendation system is designed to personalize suggestions such as videos, articles, products, services, or digital experiences based on individual user behavior, historical interactions, and preferences. The content recommendation system commonly consider that a single user is operating a given device at any point in time. A user profile of the single user is typically constructed by tracking past interactions, such as viewed content, search queries, clicks, purchases, or other engagement signals of the user on the device.

[0003] For example, in a case in which a user frequently browses tech-related articles or adds electronics to a shopping cart, the content recommendation system interprets these patterns as preferences and prioritizes similar content or items in future recommendations. In shared environments, when multiple users interact with the system together, such as during browsing, shopping, or co-viewing, the content recommendation system treats the activity as a collective interaction without distinguishing individual roles or intentions.

[0004] However, several problems arise in these systems. First, when multiple users consume content jointly, the content is often assigned equal weight to all users, regardless of the actual level of interest or engagement of each individual. Second, users may have interests they are not willing to disclose in the users group, yet current systems do not provide a mechanism to filter or limit such interests from influencing group-based recommendations.

[0005] The present disclosure is directed to overcome one or more limitations stated above or any other limitations associated with the prior art.

[0006] The information disclosed in this background of the disclosure section is only for enhancement of understanding of the general background of the disclosure and may not be taken as an acknowledgement or any form of suggestion that this information forms the prior art already known to a person skilled in the art.

[0007] The foregoing summary is illustrative only and is not intended to be in any way limiting. In addition to the illustrative aspects, embodiment, and features described above, further aspects, embodiment, and features will become apparent by reference to the drawings and the following detailed description.

[0008] In an embodiment, a method, performed by an electronic device, for generating content recommendations for a group of users, may be provided. The method may include obtaining a group knowledge graph for a group of users based on a knowledge graph of each user in the group. The knowledge graph of each user may be associated with respective user activity performed on one or more devices. The method may include obtaining, for each user in the group, a connectivity score vector based on the group knowledge graph. The connectivity score may indicate contextual contribution of the user in content recommendations with respect to other users in the group. The method may include obtaining a group connectivity score vector based on the connectivity score vectors of each user in the group of users. The method may include obtaining, based on the group connectivity score vector and the group knowledge graph, a group profile representing collective content preference of the group. The method may include recommending content to the group based on the group profile embedding.

[0009] In an embodiment, an electronic device may comprise memory storing instructions, and one or more processing unit operatively coupled to the memory and comprising processing circuitry. The one or more processing unit may individually or collectively execute the instructions to cause the electronic device to obtain a group knowledge graph for a group based on a knowledge graph of each user in the group, wherein the knowledge graph of each user is associated with respective user activity performed on one or more devices. The one or more processing unit may individually or collectively execute the instructions to cause the electronic device to obtain, for each user in the group, a connectivity score vector based on the group knowledge graph, wherein the connectivity score indicates contextual contribution of the user in content recommendations with respect to other users in the group. The one or more processing unit may individually or collectively execute the instructions to cause the electronic device to obtain a group connectivity score vector based on the connectivity score vectors of each user in the group. The one or more processing unit may individually or collectively execute the instructions to cause the electronic device to obtain, based on the group connectivity score vector and the group knowledge graph, a group profile representing collective content preference of the group. The one or more processing unit may individually or collectively execute the instructions to cause the electronic device to recommend content to the group based on the group profile.

[0010] In an embodiment, a computer-readable medium including instructions may be provided. instructions, when executed by at least one processing unit individually or collectively, may cause the electronic device to perform a method of any one of claims 1 to 8.

[0011] The accompanying drawings, which are incorporated in and constitute a part of this disclosure, illustrate exemplary embodiments and, together with the description, serve to explain the disclosed principles. The same numbers are used throughout the figures to reference like features and components. An embodiment of at least one of device and methods in accordance with an embodiment of the present subject matter are now described, by way of example only, and with reference to the accompanying figures, in which:

[0012] FIG. 1 illustrates an environment architecture of a system for generating content recommendations for a group of users, in accordance with an embodiment of the present disclosure;

[0013] FIG. 2 illustrates a block diagram of a system for generating content recommendations for a group of users, in accordance with an embodiment of the present disclosure;

[0014] FIG. 3 illustrates a process of generation of the group knowledge graph, in accordance with an embodiment of the present disclosure;

[0015] FIG. 4 illustrates a process of generation of group profile embedding, in accordance with an embodiment of the present disclosure;

[0016] FIG. 5 illustrates a process of updating the recommending content based on feedback response of users, in accordance with an embodiment of the present disclosure;

[0017] FIG. 6 illustrates a flowchart for a method for generating content recommendations for a group of users, in accordance with an embodiment of the present disclosure.

[0018] It may be appreciated by those skilled in the art that any block diagrams herein represent conceptual views of illustrative systems embodying the principles of the present subject matter. Similarly, it will be appreciated that any flow charts, flow diagrams, state transition diagrams, pseudo code, and the like represent various processes which may be substantially represented in computer readable medium and executed by a computer or processor, whether or not such computer or processor is explicitly shown.

[0019] In the present document, the word "exemplary" may be used herein to mean "serving as an example, instance, or illustration". Any embodiment or implementation of the present subject matter described herein as "exemplary" is not necessarily to be construed as preferred or advantageous over other embodiments.

[0020] While the disclosure is susceptible to various modifications and alternative forms, an embodiment thereof have been shown by way of example in the drawings and will be described in detail below. It can be understood, however, that it is not intended to limit the disclosure to the particular forms disclosed, but on the contrary, the disclosure is to cover a plurality of modifications, equivalents, and alternative falling within the spirit and the scope of the disclosure.

[0021] The terms "comprise", "comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a setup, device, or method that comprises a list of components or operations does not include only those components or operations but may include other components or operations not expressly listed or inherent to such setup or device or method. In other words, one or more elements in a device or system or apparatus proceeded by "comprises... a" does not, without more constraints, preclude the existence of other elements or additional elements in the device or system or apparatus.

[0022] In the following detailed description of an embodiment of the disclosure, reference may be made to the accompanying drawings that form a part thereof, and in which are shown by way of illustration an embodiment in which the disclosure may be practiced. An embodiment is described in sufficient detail to enable those skilled in the art to practice the disclosure, and it is to be understood that other embodiments may be utilized and that changes may be made without departing from the scope of the present disclosure. The following description is, therefore, not to be taken in a limiting sense.

[0023] The pre-trained Artificial Intelligence (AI) model may include one of: a classifier or a regression or a clustering model, which may be trained to perform a process of generating content recommendations. The pre-trained AI model may be defined by its hyper-parameters, for example, number of weights, cost function, input size, number of layers, and the like. The parameters of the pre-trained AI model may be tuned, and weights may be updated to move towards a global minimum of a cost function for the pre-trained AI model. After several epochs of the training on the feature information in the training dataset, the pre-trained AI model may be trained to output a prediction / classification result for a set of inputs. The prediction result may be indicative of a class label for each input of the set of inputs (e.g., input features extracted from new / unseen instances).

[0024] At least one processor according to the present disclosure may include at least one of a CPU (Central Processing Unit), GPU (Graphics Processing Unit), APU (Accelerated Processing Unit), MIC (Many Integrated Core), DSP (Digital Signal Processor), and NPU (Neural Processing Unit). The at least one processor may be implemented in the form of an integrated system on chip (SoC) including one or more electronic components. Each of the at least one processor may also be implemented as separate hardware (H / W).

[0025] When a method according to an embodiment of the present disclosure includes a plurality of operations, the plurality of operations may be performed by one processor or by a plurality of processors. For example, when a first operation, a second operation, and a third operation are performed by the method according to an embodiment, all of the first, second, and third operations may be performed by a first processor, or the first and second operations may be performed by a first processor (for example, a general-purpose processor), and the third operation may be performed by a second processor (for example, a processor dedicated to artificial intelligence). However, the embodiments of the present disclosure are not limited thereto.

[0026] The at least one processor according to the present disclosure may be implemented as a single core processor or a multicore processor.

[0027] When a method according to an embodiment of the present disclosure includes a plurality of operations, the plurality of operations may be performed by one core or by a plurality of cores included in the one or more processors. FIG. 1 illustrates an environment architecture 100 for generating content recommendations for a group of users, in accordance with an embodiment of the present disclosure. The environment architecture 100 may include a device 102, a server 106, and an electronic device 108. The embodiment of the environment architecture 100 shown in FIG. 1 is for illustration only. Other embodiments of the environment architecture 100 could be used without departing from the scope of this disclosure. The environment architecture 100 may be implemented by more constituent elements than the constituent elements illustrated in FIG. 1, but the same are not explained for the sake of brevity.

[0028] In an embodiment, the device 102 may refer to any electronic device that is capable of displaying, playing, or otherwise delivering content or recommendations to a user or group of users. The device 102 may include, but is not limited to, smart TVs, tablets, smartphones, laptops, desktops, voice assistants with screens, or other interactive display units found in smart home or IoT environments. The device 102 may serve as the device on which content recommendations are presented (e.g., the platform where personalized or group-based content suggestions are presented to a group of users 104).

[0029] In an embodiment, the server 106 may be configured to store and manage device-level knowledge graphs. Each of the device-level knowledge graphs may correspond to the interactions and behaviors of users recorded on a specific device. The device-level knowledge graph may be structured as directed or undirected graphs comprising nodes and edges. Each node may represent an entity associated with an activity on the device, such as content items, users, applications, or contextual factors (e.g., time, location, or device metadata). Edges may represent the semantic or behavioral relationships between these entities, including, but not limited to, interactions such as content consumption, communication events, application usage, or user co-presence.

[0030] In an embodiment, the generation of the device-level knowledge graph may be performed by the server 106. In an embodiment, the generation of the device-level knowledge graph may be performed by the electronic device 108.

[0031] To generate the device-level knowledge graphs, the electronic device 108 or server 106 may collect activity data corresponding to user interactions occurring on one or more devices. These interactions may include, but are not limited to, watching videos, sending messages, accessing applications, or browsing content. In a non-limiting example of the present disclosure, if a first user sends a message to a second user using a messaging application on their smartphone, the device-level knowledge graph may record this event by including nodes representing the first user, the second user, and the messaging application. The edge between the first user and the second user may be labeled "messaged," indicating the communication interaction. In a non-limiting example of the present disclosure, a shared device such as a smart TV may be used by multiple users. When a first user and a third user watch a movie together, the smart TV may generate a device-level knowledge graph that includes nodes for the first user, the third user, the movie content. An edge labeled "co-watched" may be created between the user nodes and the content node, capturing the joint interaction. In an embodiment, additional metadata, such as the time of viewing, content genre, or resolution, may be appended.

[0032] In an embodiment, the server 106 may be implemented as a cloud server and may execute operations through web applications, cloud applications, HTTP requests, repository operations, file transfer, or the like. An example implementation of the server 106 may include, but are not limited to, a database server, a file server, a web server, a media server, an application server, a mainframe server, or a cloud computing server. In an embodiment, the server 106 may be implemented as a plurality of distributed cloud-based resources by use of several technologies that are well known to those ordinarily skilled in the art. A person with ordinary skill in the art will understand that the scope of the disclosure may not be limited to the implementation of the server 106 and the electronic device 108 as two separate entities. In an embodiment, the functionalities of the server 106 may be incorporated in its entirety or at least partially in the electronic device 108, without a departure from the scope of the disclosure.

[0033] In an embodiment, the electronic device 108 may communicate with the device 102 and server 106 of the architecture 100 via wireless communication network, and may be configured to generate content recommendations for a group of users 104. In an embodiment, the electronic device 108 may include memory 110 and the processing unit 112. The detailed functioning of the electronic device 108 is further explained in FIG. 2 in forthcoming paragraphs of the present disclosure. A person with ordinary skill in the art will understand that the scope of the disclosure may not be limited to the implementation of the electronic device 108 and the device 102 as two separate entities. In an embodiment, the functionalities of the electronic device 108 can be incorporated in its entirety or at least partially in the device 102, without a departure from the scope of the disclosure.

[0034] FIG. 2 illustrates a block diagram of an electronic device 200 for generating content recommendations for a group of users, in accordance with an embodiment. FIG. 2 is explained in conjunction with FIG. 1. In an embodiment, the electronic device 200 (same as the electronic device 108 of FIG. 1) may include memory 202, a processing unit 204, and communication interface 206. All components of the electronic device 200 may be communicably coupled with each other, directly or indirectly.

[0035] The memory 202 may include suitable logic, circuitry, or interfaces that may be configured to store the one or more instructions to be executed by the processing unit 204. In an embodiment, the memory 202 may store one or more instructions that, when executed by the processing unit 204, cause the electronic device 200 to train a pre-trained artificial intelligence (AI) model. In an example, the memory 202 may represent any type of non-transitory computer readable medium such as random-access memory (RAM), read only memory (ROM), magnetic disk or tape, optical disk, flash memory, or holographic memory. In an embodiment, the memory 202 may include a combination of random-access memory and read only memory and may include data / instructions related to processing of one or more components of the electronic device 200.

[0036] It is to be appreciated that the processing unit 204 may execute a pre-trained AI model that is configured to perform the process of generating content recommendation. The pre-trained AI model may be represented by electronic data, which may be implemented as, for example, a software component of an application executable on the content recommendation system. The pre-trained AI model may rely on libraries, external scripts, or other instructions for execution by a processing device, such as a controller. The pre-trained AI model may include code and routines configured to enable a computing device, such as the processing unit to perform one or more operations. In an embodiment, the pre-trained AI model may be implemented using hardware including a processor, a field-programmable gate array (FPGA), or an application-specific integrated circuit (ASIC). In an embodiment, the pre-trained AI model may be implemented using a combination of hardware and software. Further details of the pre-trained AI model have been omitted from the disclosure for the sake of brevity.

[0037] In an embodiment, the device 102 may be configured to support recommendation functionality may initiate a target audience determination process in response to detection of a recommendation-related activity. The recommendation-related activity may include, but is not limited to, activation of the device 102, navigation to a homepage or dashboard interface, execution of a content-related application, or initiation of a user action indicative of multi-user engagement. The recommendation-related activity may trigger a process for determining group of users constituting the target audience for a current recommendation process. It may be appreciated that triggering process may be initiated not only by the group of users but also by a single user. As used herein, the term "group" may refer to a group of users, which constitutes the target audience, unless otherwise specified.

[0038] In an embodiment, the electronic device 200 may be configured to communicably coupled with the device 102 and may be configured to determine the target audience (e.g., group of users) by executing one or more user detection procedures. In an embodiment, the device 102 may include one or more image-capturing components, including, but not limited to, cameras. The one or more image-capturing components may be configured to acquire visual data from environment surrounding the device 102. The acquired image data may be processed, either by the device 102 or by the electronic device 200, to identify each user in the group of users present within a defined proximity to the device 102. In an embodiment, the device 102 may include one or more audio input components, including, but not limited to, microphones. The one or more audio input components may be configured to capture speech or other audio signals of users in the group. The captured audio input may be processed, either by the device 102 or by the electronic device 200, to identify users based on stored voice signatures. In an embodiment, the device 102 may be configured to be operatively and / or communicatively coupled with one or more devices in a local Internet of Things (IoT) environment, including, but not limited to, one or more smart sensors, presence detectors, or other contextual signal sources. The device 102 may analyze signals received from such IoT components to identify the users. In an embodiment, the device 102 may receive the signals from the IoT components and transmit the received signals to the electronic device 200. The electronic device 200 may be configured to identify the group based on the received signals. In an embodiment, the device 102 may perform wireless signal scanning to detect active wireless signals (e.g., Wi-Fi or Bluetooth) emitted by devices associated with the users. Upon detecting such signals, the device 102 or the electronic device 200 may identify users proximate to the device 102. The foregoing detection techniques may be performed individually or in combination to determine the group of users.

[0039] In an embodiment, upon determination of the target audience (e.g., the group), the communication interface 206, may receive a list of devices associated with each of the group of users. In an embodiment, the communication interface 206 may receive a list of devices from the server 106. In an embodiment, the list of devices may be stored in the memory 202.

[0040] In an embodiment, the communication interface 206 may receive device-level knowledge graph. In an embodiment, the device-level knowledge graph may be stored in the memory 202. In an embodiment, based on the list of devices and the device-level knowledge graph, the processing unit 204 may cause the electronic device 200 to perform a traversal operation on the device-level knowledge graph to extract a subgraph comprising nodes within N-hops of a user node.

[0041] In an embodiment, a group knowledge graph generation unit 208 of the processing unit 204 may cause the electronic device 200 to perform a process of scaling relations using a connectivity score on the subgraph. The group knowledge graph generation unit 208 may obtain (e.g., compute) the connectivity score based on users and interactions represented within the subgraph. The group knowledge graph generation unit 208 may apply connectivity score to each node and relation in the subgraph. This operation of applying the score may include modifying or weighting the relevance or contribution of at least a subset of nodes and / or relations based on the corresponding connectivity score. The connectivity score may correspond to contextual contribution of the user in content recommendations with respect to the group (e.g., with respect to other users in the group). The connectivity score may reflect the influence or significance of the associated user or interaction in group settings, and may be derived based on prior observations of shared device usage or historical interaction behavior. In a non-limiting example of the present disclosure, in a scenario where multiple users use the same device either individually or together, the electronic device 200 may use the connectivity score to differentially weight the contributions of content nodes. A content item viewed collaboratively by multiple users may be weighted differently compared to the same content viewed individually.

[0042] In an embodiment, the scaled subgraph may constitute a partial knowledge graph associated with the corresponding user and device. This process may be repeated for each device in the list of devices associated with the user, resulting in a set of partial knowledge graphs.

[0043] In an embodiment, the group knowledge graph generation unit 208 may cause the electronic device 200 to generate a group knowledge graph for a group of users based on a partial knowledge graphs of each user in the group of users.

[0044] FIG. 3 illustrates a process of generation of the group knowledge graph. As shown in FIG. 3, based on partial knowledge graphs for each user, the group knowledge graph generation unit 208 may cause the electronic device 200 to obtain a user-wise knowledge graph for each user in the target audience (e.g., the group of users). In an embodiment, the group knowledge graph generation unit 208 may combine (e.g., merge) at least a subset of the partial knowledge graphs 302 to obtain user-wise knowledge graphs 304. In an embodiment, the group knowledge graph generation unit 208 may identify and align common nodes and semantically equivalent edges across at least a subset of the partial knowledge graphs 302.

[0045] In an embodiment, the group knowledge graph generation unit 208 may generate a group knowledge graph 306 by combining the user-wise knowledge graphs 304. In an embodiment, the group knowledge graph generation unit 208 may align at least a subset of nodes and / or edges across the user-wise knowledge graphs 304. In an embodiment, the group knowledge graph generation unit 208 may perform the alignment based on shared content interests and social relationships.

[0046] In a non-limiting example of the present disclosure, a first user and a second user may be identified as users of the group of users. Each user may have interacted with multiple devices in the past, and each device may maintain a partial knowledge graph indicative of user-specific activities. A partial knowledge graph of the first user associated with a first device may include a node representing the content item "Cricket Highlights" connected with a user node, via a "watched" edge. A partial knowledge graph of the first user associated with the second device may include a node representing "Cricket live App" and a "used" edge. The electronic device may merge the partial knowledge graphs by identifying common or semantically linked nodes, such as both nodes relating to the domain of sports, and may combine them into a user-wise knowledge graph for the first user.

[0047] A similar process may be performed for the second user, whose partial knowledge graphs including nodes such as "Badminton Final" and "YouTube," both linked to the user node via content consumption or application usage relations. The partial knowledge graphs may be merged into a user-wise knowledge graph of the second user by aligning nodes that share categorical or contextual meaning (e.g., both associated with video-based sports content).

[0048] In an embodiment, the group knowledge graph generation unit 208 may construct a group knowledge graph by merging the user-wise knowledge graphs of the first user and the second user. In a case in which both users have consumed content categorized under "Sports," a higher-level node such as "Sports" may be introduced to semantically unify the user graphs.

[0049] In an embodiment, a connectivity and exposable interest unit 210 of the processing unit 204 may cause the electronic device 200 to generate an exposable interest embedding and a connectivity score vector for each user. In the subsequent paragraphs, the process for generating the connectivity score vector is described first, followed by a detailed description of the generation of the exposable interest embedding for each user in the target audience (e.g., the group).

[0050] In an embodiment, a connectivity score vector may be generated for each user in the group to represent the user-relative influence in group-based content recommendation. The connectivity and exposable interest unit 210 may cause the electronic device to receive the group knowledge graph 306 from the group knowledge graph generation unit 208. The group knowledge graph 306 may include nodes representing users and edges representing various types of relationships. The group knowledge graph 306 may include representations of user-user interactions and user-content engagements. The connectivity and exposable interest unit 210 may cause the electronic device to prune the group knowledge graph 306 to retain edges that represent user-user relations within the group of users. This pruning operation may preserve edges representing user-user relations that signify prior co-engagements or behavioral similarities, such as shared viewing history, social proximity, or repeated co-participation in content consumption events. Based on the pruning process, the connectivity and exposable interest unit 210 may cause the electronic device 200 to obtain a group knowledge subgraph, which may include the users in the group and the edges that connect the users based on relevant behavioral or social context. The group knowledge subgraph may include a first pruned group knowledge graph.

[0051] In an embodiment, the connectivity and exposable interest unit 210 may process the group knowledge subgraph using a graph-based neural network to obtain (e.g., compute) a node embedding for each user node. The graph-based neural network may include the first graph-based neural network. The graph-based neural network may be, but is not limited to, a Graph Convolutional Network (GCN), Graph Attention Network (GAT), or Temporal Graph Network (TGN).

[0052] The node embedding may be transformed, using one or more fully connected layers or projection functions, to generate a connectivity score vector for each user. The connectivity score vector may represent the degree of influence or importance of the user within the group for the purpose of content recommendation. Users who are more centrally connected or who exhibit stronger relational patterns with other users in the group may be assigned higher connectivity scores, which influence the content selection process accordingly.

[0053] In a non-limiting example of the present disclosure, a group of friends may be streaming music together, such as during a party or celebration. In this group, the music enthusiast user (e.g., someone who is highly passionate about music) may have a higher connectivity score. This is because the music enthusiast user's preferences are more likely to align with the general direction of the group interests, and the music enthusiast user may be more active in influencing the playlist or suggesting tracks.

[0054] In an embodiment, the connectivity and exposable interest unit 210 may cause the electronic device 200 to prune the group knowledge graph 306 based on user-content relationships within the group. In an embodiment, the connectivity and exposable interest unit 210 may evaluate, using a graph-based neural network, content preferences of each user relevant to exposure to the group based on the pruned knowledge graph. The graph-based neural network may include a second graph-based neural network. The pruned knowledge graph may include a second pruned knowledge graph. In an embodiment, the connectivity and exposable interest unit 210 may obtain (e.g., generate) exposable interest embedding based on the evaluation.

[0055] In an embodiment, the connectivity and exposable interest unit 210 may cause the electronic device to prune the group knowledge graph 306 by retaining edges representing user-content relations within the group to obtain (e.g., generate) a pruned knowledge graph. The pruned knowledge graph (e.g., the second pruned knowledge graph) may be passed through one or more graph neural network layers to obtain (e.g., compute) node embedding that represent the user's behavior in the context of content consumption within the group of users. The node embedding for each user may indicate the set of content types that the user is likely to engage with or prefer to share when interacting with the group of users. The node embedding may not depend on common interests among group members, but, based on historical patterns in similar contexts, may reflect content that the user is comfortable pursuing in the presence of the group.

[0056] For example, a user may have previously watched romantic movies, anime, documentaries, and news debates. When this user is placed in a group with a parent or a colleague, the connectivity and exposable interest unit 210 may examine past content consumption behavior involving the parent or the colleague. The historical data may show that the user has only watched documentaries and news debates with the parent, and has only discussed or shared news-related content with the colleague. Romantic movies and anime may have only been consumed by the user alone or with close friends, and never in the presence of the parent or the colleague.

[0057] In an embodiment, a target audience profile unit 212 of the processing unit 204 may cause the electronic device 200 to obtain (e.g., generate) group profile representing collective content preference of the group of users. The target audience profile unit 212 may encode the group profile to obtain a group profile embedding. Now moving towards FIG. 4 in order to understand the process related to generation of group profile embedding.

[0058] FIG. 4 illustrates a process of generation of group profile embedding, in accordance with an embodiment of the present disclosure. The group profile unit 212 may cause the electronic device 200 to obtain (e.g., receive) the group knowledge graph 306 and exposable interest embedding 402 of each user in the group of users.

[0059] In an embodiment, the group profile unit 212 may cause the electronic device 200 to obtain, for each user, a set of candidate content nodes, in the group knowledge graph 306, associated with the user. For each candidate content node, the content node embedding may be obtained. The content embedding may be weighted using the exposable interest embedding 402 of the corresponding user. The weighting the content embedding may indicate the relevance of the content with the user's exposable interests. To determine the relevance, the group profile unit 212 may compare, for each candidate content node, the content embedding associated with the node to the exposable interest embedding 402 of the corresponding user. The comparison may be performed using a similarity function, such as cosine similarity, or by applying a learned neural network model. Based on the comparison, the relevance of the content with the user's exposable interests may be determined. The group profile unit 212 may insert a group node representing the collective target audience (e.g., the group), and connect the weighted candidate content nodes to the group node to obtain (e.g., construct) the final group knowledge graph 404.

[0060] In an embodiment, the group profile unit 212 may generate a group connectivity score vector based on the connectivity score 406 of each user in the group. In an embodiment, the group profile unit 212 may generate a group connectivity score vector by aggregating the connectivity scores 406 of each user in the group. The aggregation may be performed using one or more techniques including, but not limited to, averaging, weighted averaging, dense neural layers, or attention-based mechanisms.

[0061] The group profile unit 212 may cause the electronic device to utilize the final group knowledge graph 404 and the group connectivity score vector to generate a group profile. In an embodiment, group profile embedding and group generative assets preferences may be obtained based on the group profile.

[0062] In an embodiment, the group profile unit 212 may cause the electronic device 200 to generate the group profile embedding and the group generative assets preferences using a language model. The language model may include an auto-regressive language model. The group profile unit 212 may process the final group knowledge graph 404, which includes user-content relations and hierarchical relations. The user-content relations may be obtained (e.g., extracted) and provided to the language model, with the group connectivity score vector to prioritize the importance of different user-content relations. In an embodiment, the group profile unit 212 may cause the electronic device 200 to generate a group profile by providing, as input to the language model, the metadata obtained from the final group knowledge graph 404, which includes the hierarchical relations and the user-content relations. The group profile may include the audience profile and the group generative assets preferences. The group profile may be encoded using a language encoder to generate the group profile embedding.

[0063] In an embodiment, the group profile unit 212 may cause the electronic device 200 to generate group profile embedding and group generative assets preferences by aggregating content embeddings from the final group knowledge graph 404 and weighting the content embeddings based on connectivity scores 406 of each user in the group. The group profile unit 212 may cause the electronic device to obtain the content embeddings from the final group knowledge graph 404 and perform a scaling operation using the connectivity scores 406 for each content node. The scaling operation may adjust the importance of each content node based on the connectivity score 406 of the corresponding user, prioritizing content that is more relevant or impactful for the group. The weighted content embeddings may be aggregated to form a group profile embedding. In an embodiment, the group generative assets preferences may be obtained by applying similar weighting principles to content embeddings related to specific assets.

[0064] In an embodiment, a recommendation unit 214 of the processing unit 204 may cause the electronic device 200 to perform a content recommendation process. In this process, the recommendation unit 214 may cause the electronic device 200 to obtain (e.g., extract) a plurality of contents for recommendation by communicating with the device 102. The plurality of contents may be not limited to mobile-based content such as multimedia content, food-related content, or other application-specific digital content. The recommendation unit 214 may cause the electronic device 200 to encode the plurality of contents using a content encoder to generate content embeddings. The content encoder may include, not limited to, bidirectional encoder representations from transformers (BERT), robustly optimized BERT pretraining approach (RoBERTa), or unified language model (UniLM). Upon the generation of the content embeddings , the recommendation unit 214 may perform scoring procedure to compute the similarity between the content embeddings and the group profile embedding. The scoring procedure may include, but is not limited to, cosine similarity, dot product, pearson correlation, or manhattan distance. Based on the computed similarity, the recommendation unit 214 may cause the electronic device 200 to determine a rank for at least a subset of the plurality of contents. In an embodiment, the recommendation unit 214 may perform sorting based on the determined rank. In an embodiment, the recommendation unit 214 may generate a ranked list base on the determined rank.

[0065] Upon the generation of the ranked list, recommendation unit 214 may cause the electronic device 200 to obtain (e.g., extract) multimedia data from the ranked plurality of contents. The multimedia data may include audio, text, or image frames, based on the type of content. The multimedia data may be processed by the recommendation unit 214 in conjunction with multimedia preferences derived from the group generative assets preferences to generate multimedia assets. The generative assets preferences may be used as input prompts that guide how the extracted multimedia data should be transformed or adapted into personalized generative multimedia assets.

[0066] Upon the generation of multimedia assets, the recommendation unit 214 may cause the electronic device 200 to encode multimedia assets to generate encoded multimedia assets. In an embodiment, the recommendation unit 214 may cause the electronic device 200 to aggregate exposable interest embedding 402 of each user in the group to generate aggregated exposable interest embedding 402. In an embodiment, the recommendation unit 214 may cause the electronic device 200 to compute a contrastive loss between the encoded multimedia assets and aggregated exposable interest embedding. Based on the computed contrastive loss, at least one multimedia asset may be selected from the generated multimedia assets for at least a subset of the contents in the ranked list. In an embodiment, at least one content may be recommended to the group of users based on the selected multimedia assets and the ranked plurality of contents.

[0067] Now moving towards FIG. 5 to understand the feedback mechanism of the electronic device 200. FIG. 5 illustrates a process of updating the recommending content based on feedback response of users. In an embodiment, the electronic device 200 may recommend a list of contents to a group of users on the device 102. After viewing the recommendations, the users may respond by consuming some items of at least one content. A feedback unit 216 of the processing unit 204 may cause the electronic device 200 to collect the responses. Based on the responses, the feedback unit 216 may compute a loss function, which is used to update and further train various components of the processing unit 204 to generate the updated recommendation content.

[0068] In a non-limiting example of the present disclosure, the electronic device 200 may recommend videos to a group of users on their mobile phones. The videos may include topics like cooking, comedy, fitness tips, or travel. After viewing the recommendations, the users may respond by watching some videos fully, skipping others, and liking or disliking a few. The recommendation unit 214 may collect the responses. The recommendation unit 214 may detect that most users skipped the travel videos after just a few seconds. At the same time, cooking videos may have been watched until the end and may have received the most likes. A few users may have shared the cooking videos with others. Based on the feedback, the recommendation unit 214 may reduce the number of travel videos in future recommendations. Instead, the recommendation unit 214 may increase the number of cooking and recipe-related videos for the group. In the next recommendation cycle, the users may see more food-related content, which matched their preferences better.

[0069] Further, the following use cases illustrate various applications of the invention.

[0070] In a non-limiting example of the present disclosure, when a user intends to share content such as an image, video, or other multimedia via a messaging application, the electronic device 200 may be configured to suggest relevant contacts or groups with whom to share the content. To enable more contextually relevant sharing suggestions, the electronic device 200 may leverage exposable interest embedding 402. The exposable interest embedding 402 may be computed for each contact or group and represent the degree to which a given content item aligns with the interests of the potential recipient, based on historical interactions and behavioral signals. For example, when the user selects a food-related video for sharing, the electronic device 200 may evaluate the exposable interest embedding 402 and recommend the contacts or groups such as family members or friends who have previously shown engagement with food-related content.

[0071] In a non-limiting example of the present disclosure, the electronic device 200 may be configured to provide group-based food recommendations in a family context involving, for instance, a mother, a father, and a child. The electronic device 200 may analyze individual user's preferences, such as the child's preference for snacks, the mother's preference for healthy food, and the father's preference for spicy dishes. By applying exposable interest embedding 402 associated with each user, the electronic device 200 may prioritize food items that reflect the group's collective interests while still accommodating individual preferences. For example, the electronic device 200 may recommend a meal that includes healthy components, snack options, and spicy elements, thereby catering to the varied tastes within the family. The system may utilize connectivity scores 406 to further refine the recommendations, assigning higher priority to food items with greater relevance to specific user in the group.

[0072] FIG. 6 is a flowchart of a method 600 for generating content recommendations for a group of users. FIG. 6 is described in conjunction with reference from FIGs. 1-5. With reference to FIG. 6, there is shown a flowchart 600 that may correspond to a method for generating content recommendations for a group of users. Further, for ease of explanation, in an embodiment described below, the method may be implemented by an operator or a manufacturer, as described with reference to FIGs. 2. The method illustrated in the flowchart may start from operation 602.

[0073] The method may start at operation 602, at operation 602 the method 600 may include obtaining (e.g., generating) a group knowledge graph for a group of users based on a knowledge graph of each user in the group. The knowledge graph of each user may be associated with respective user activity performed on one or more devices. In an embodiment, the electronic device 200 of FIG. 2 may be configured to perform the operations disclosed in operation 602.

[0074] At operation 604, the method 600 may include obtaining (e.g., generating), for each user in the group, a connectivity score vector based on the group knowledge graph. The connectivity score may indicate contextual contribution of the user in content recommendations with respect to other users in the group. In an embodiment, the electronic device 200 of FIG. 2 may be configured to perform the operations disclosed in operation 604.

[0075] At operation 606, the method 600 may include obtaining a group connectivity score vector base on the connectivity score vectors of each user in the group. In an embodiment, the method may include aggregating the connectivity score vectors of each user in the group. In an embodiment, the electronic device 200 of FIG. 2 may be configured to perform the operations disclosed in operation 606.

[0076] At operation 608, the method 600 may include obtaining (e.g., generating), based on the group connectivity score vector and the group knowledge graph, a group profile representing collective content preference of the group. In an embodiment, the electronic device 200 of FIG. 2 may be configured to perform the operations disclosed in operation 608.

[0077] At operation 610, the method 600 may include recommending content to the group of users based on the group profile. In an embodiment, the electronic device 200 of FIG. 2 may be configured to perform the operations disclosed in operation 610.

[0078] The order in which the flowchart 600 is described is not intended to be construed as a limitation, and any number of the described method blocks may be combined in any order to implement the flowchart 600 or alternate methods. Additionally, individual blocks may be deleted from the flowchart 600 without departing from the spirit and scope of the subject matter described herein. Furthermore, the method may be implemented in any suitable hardware, software, firmware, or combination thereof.

[0079] In an embodiment, the method 600 may include, for generating the group knowledge graph, detecting the group of users for which content is to be recommended, obtaining (e.g., extracting) the knowledge graph of each user in the group from a database including knowledge graphs of a plurality of users, identifying overlapping features among the knowledge graphs of the users in the group, and obtaining (e.g., generating) the group knowledge graph by integrating the knowledge graphs of the users based on the overlapping features.

[0080] In an embodiment, the method 600 may include, for generating the connectivity score vector, obtaining a first pruned group knowledge graph by pruning the group knowledge graph based on user-user relationships and obtaining (e.g., generating) the connectivity score vector by processing the first pruned group knowledge graph using a first graph-based neural network.

[0081] In an embodiment, the method 600 may include, for obtaining (e.g., generating) the group profile, obtaining (e.g., generating), for each user in the group, exposable interest embedding based on the group knowledge graph. the exposable interest embedding may indicate the respective user's content preferences exposable to the group, modifying the group knowledge graph based on the exposable interest embedding to obtain (e.g., generate) a final group knowledge graph, and obtaining (e.g., generating), based on the group connectivity score vector and the final group knowledge graph, the group profile.

[0082] In an embodiment, the method 600 may include obtaining a second pruned knowledge graph by pruning the group knowledge graph based on user-content relationships, evaluating, using a second graph-based neural network, content preferences of each user relevant to exposure to the group, based on the second pruned knowledge graph, and obtaining (e.g., generating) exposable interest embedding based on the evaluation.

[0083] In an embodiment, the method 600 may include, for recommending content to the group, encoding the group profile to obtain a group profile embedding, obtaining (e.g., extracting) a plurality of contents for recommendation, determining a rank for each of the plurality of contents based on the group profile embedding; and recommending content to the group based on the rank for each of the plurality of contents.

[0084] In an embodiment, the method 600 may include, for recommending content to the group based on the rank for each of the plurality of contents, obtaining (e.g., extracting) multimedia data from the plurality of contents, obtaining (e.g., receiving) a multimedia preferences based on the group profile, obtaining (e.g., generating) candidate multimedia assets based on the multi-media preferences and the multimedia data, computing a contrastive loss between the candidate multimedia assets and exposable interest embedding (e.g., aggregated exposable interest embedding), indicating the respective user's content preferences exposable to the group, selecting at least one multimedia asset from the candidate multimedia assets for at least the a subset of the plurality of contents based on the contrastive loss, and recommending content to the group based on the at least one multimedia asset and the rank for each of the plurality of contents.

[0085] In an embodiment, the method 600 may include updating the knowledge graph of each user based on response from the group.

[0086] The disclosed method with reference to FIG. 6, or one or more operations of the electronic device 200 explained with reference to FIG. 2 may be implemented using software including computer-executable instructions stored on one or more computer-readable media (e.g., non-transitory computer-readable media, such as one or more optical media discs, volatile memory components (e.g., DRAM or SRAM), or non-volatile memory or storage components (e.g., hard drives or solid-state non-volatile memory components, such as Flash memory components) and executed on a computer (e.g., any suitable computer, such as a laptop computer, net book, Web book, tablet computing device, smart phone, or other mobile computing device). Such software may be executed, for example, on a single local computer.

[0087] In an embodiment, an electronic device may comprise memory (202) storing instructions, and one or more processing unit (204) operatively coupled to the memory (202) and comprising processing circuitry. The one or more processing unit (204) may individually or collectively execute the instructions to cause the electronic device (200) to obtain a group knowledge graph for a group based on a knowledge graph of each user in the group, wherein the knowledge graph of each user is associated with respective user activity performed on one or more devices. The one or more processing unit (204) may individually or collectively execute the instructions to cause the electronic device (200) to obtain, for each user in the group, a connectivity score vector based on the group knowledge graph, wherein the connectivity score indicates contextual contribution of the user in content recommendations with respect to other users in the group. The one or more processing unit (204) may individually or collectively execute the instructions to cause the electronic device (200) to obtain a group connectivity score vector based on the connectivity score vectors of each user in the group. The one or more processing unit (204) may individually or collectively execute the instructions to cause the electronic device (200) to obtain, based on the group connectivity score vector and the group knowledge graph, a group profile representing collective content preference of the group. The one or more processing unit (204) may individually or collectively execute the instructions to cause the electronic device (200) to recommend content to the group based on the group profile.

[0088] In an embodiment, the one or more processing unit (204) may individually or collectively execute the instructions to cause the electronic device (200) to detect the group for which content is to be recommended. The one or more processing unit (204) may individually or collectively execute the instructions to cause the electronic device (200) to obtain the knowledge graph of each user in the group from a database including knowledge graphs of a plurality of users. The one or more processing unit (204) may individually or collectively execute the instructions to cause the electronic device (200) to identify overlapping features among the knowledge graphs of the users in the group. The one or more processing unit (204) may individually or collectively execute the instructions to cause the electronic device (200) to obtain the group knowledge graph by integrating the knowledge graphs of the users based on the overlapping features.

[0089] In an embodiment, the one or more processing unit (204) may individually or collectively execute the instructions to cause the electronic device (200) to obtain a first pruned group knowledge graph by pruning the group knowledge graph based on user-user relationships. The one or more processing unit (204) may individually or collectively execute the instructions to cause the electronic device (200) to obtain the connectivity score vector by processing the first pruned group knowledge graph using a first graph-based neural network.

[0090] In an embodiment, the one or more processing unit (204) may individually or collectively execute the instructions to cause the electronic device (200) to obtain, for each user in the group, exposable interest embedding based on the group knowledge graph, wherein the exposable interest embedding indicates the respective user's content preferences exposable to the group. The one or more processing unit (204) may individually or collectively execute the instructions to cause the electronic device (200) to modify the group knowledge graph based on the exposable interest embedding to obtain a final group knowledge graph. The one or more processing unit (204) may individually or collectively execute the instructions to cause the electronic device (200) to obtain the group profile, based on the group connectivity score vector and the final group knowledge graph.

[0091] In an embodiment, the one or more processing unit (204) may individually or collectively execute the instructions to cause the electronic device (200) to obtain a second pruned knowledge graph by pruning the group knowledge graph based on user-content relationships. The one or more processing unit (204) may individually or collectively execute the instructions to cause the electronic device (200) to evaluate, using a second graph based neural network, content preferences of each user relevant to exposure to the group, based on the second pruned knowledge graph. The one or more processing unit (204) may individually or collectively execute the instructions to cause the electronic device (200) to obtain the exposable interest embedding based on the evaluation.

[0092] In an embodiment, the one or more processing unit (204) may individually or collectively execute the instructions to cause the electronic device (200) to encode the group profile to obtain a group profile embedding. The one or more processing unit (204) may individually or collectively execute the instructions to cause the electronic device (200) to obtain a plurality of contents for recommendation. The one or more processing unit (204) may individually or collectively execute the instructions to cause the electronic device (200) to determine a rank for each of the plurality of contents based on the group profile embedding. The one or more processing unit (204) may individually or collectively execute the instructions to cause the electronic device (200) to recommend content to the group based on the rank for each of the plurality of contents.

[0093] In an embodiment, the one or more processing unit (204) may individually or collectively execute the instructions to cause the electronic device (200) to obtain multimedia data from the plurality of contents. The one or more processing unit (204) may individually or collectively execute the instructions to cause the electronic device (200) to obtain a multimedia preferences based on the group profile. The one or more processing unit (204) may individually or collectively execute the instructions to cause the electronic device (200) to obtain candidate multimedia assets based on the multimedia preferences and the multimedia data. The one or more processing unit (204) may individually or collectively execute the instructions to cause the electronic device (200) to compute a contrastive loss between the candidate multimedia assets and exposable interest embedding indicating the respective user's content preferences exposable to the group. The one or more processing unit (204) may individually or collectively execute the instructions to cause the electronic device (200) to select at least one multimedia asset from the candidate multimedia assets for at least a subset of the plurality of contents based on the contrastive loss. The one or more processing unit (204) may individually or collectively execute the instructions to cause the electronic device (200) to recommend content to the group based on the at least one multimedia asset and the rank for each of the plurality of contents.

[0094] In an embodiment, the one or more processing unit (204) may individually or collectively execute the instructions to cause the electronic device (200) to update the knowledge graph of each user based on response from the group.

[0095] In an embodiment, a computer-readable medium including instructions may be provided. instructions, when executed by at least one processing unit (204) individually or collectively, may cause the electronic device (200) to perform a method of any one of claims 1 to 8.

[0096] Technical Effect:

[0097] An embodiment of the present disclosure may provide a content recommendation system that uses the interests which users are comfortable exposing within a specific group context, rather than relying on generic common interests across users.

[0098] An embodiment of the present disclosure may provide a mechanism for adjusting the weight of consumed content differently for each user in a group based on individual behavioral and contextual factors, rather than applying a uniform weight to all users, thus enabling more accurate and personalized future recommendations.

[0099] Furthermore, one or more computer-readable storage media may be utilized in implementing an embodiment consistent with the present disclosure. A computer-readable storage medium may refer to any type of physical memory on which information or data readable by a processor may be stored. Thus, a computer-readable storage medium may store instructions for execution by one or more processors, including instructions for causing the processor(s) to perform operations or stages consistent with an embodiment. The term "computer-readable medium" may be understood to include tangible items and exclude carrier waves and transient signals, i.e., be non-transitory. Examples may include Random Access Memory (RAM), Read-Only Memory (ROM), volatile memory, non-volatile memory, hard drives, CD (Compact Disc) ROMs, DVDs, flash drives, disks, or any other known physical storage media.

[0100] It will be understood by those within the art that, in general, terms used herein, and are generally intended as "open" terms (e.g., the term "including" may be interpreted as "including but not limited to," the term "having" may be interpreted as "having at least," the term "includes" may be interpreted as "includes but is not limited to," etc.). For example, as an aid to understanding, the detail description may contain usage of the introductory phrases "at least one" and "one or more" to introduce recitations. However, the use of such phrases may not be construed to imply that the introduction of a recitation by the indefinite articles "a" or "an" limits any particular part of description containing such introduced recitation to disclosure containing only one such recitation, even when the introductory phrases "one or more" or "at least one" and indefinite articles such as "a" or "an" (e.g., "a" and / or "an" may typically be interpreted to mean "at least one" or "one or more") are included in the recitations; the same holds true for the use of definite articles used to introduce such recitations. In addition, even if a specific part of the introduced description recitation is explicitly recited, those skilled in the art will recognize that such recitation may typically be interpreted to mean at least the recited number (e.g., the bare recitation of "two recitations," without other modifiers, typically means at least two recitations or two or more recitations).

[0101] While various aspects and embodiments have been disclosed herein, other aspects and embodiments will be apparent to those skilled in the art. The various aspects and embodiments disclosed herein are for purposes of illustration and are not intended to be limiting, with the true scope and spirit being indicated by the following detailed description.

Claims

1.A method, performed by an electronic device (200), for generating content recommendations for a group of users, the method comprising:obtaining a group knowledge graph (306) for a group of users based on a knowledge graph (304) of each user in the group, wherein the knowledge graph (304) of each user is associated with respective user activity performed on one or more devices;obtaining, for each user in the group, a connectivity score vector based on the group knowledge graph (306), wherein the connectivity score indicates contextual contribution of the user in content recommendations with respect to other users in the group;obtaining a group connectivity score vector based on the connectivity score vectors of each user in the group;obtaining, based on the group connectivity score vector and the group knowledge graph (306), a group profile representing collective content preference of the group; andrecommending content to the group based on the group profile.2.The method as claimed in claim 1, wherein obtaining the group knowledge graph (306) comprises:detecting the group for which content is to be recommended;obtaining the knowledge graph (304) of each user in the group from a database including knowledge graphs of a plurality of users;identifying overlapping features among the knowledge graphs (304) of the users in the group; andobtaining the group knowledge graph (306) by integrating the knowledge graphs (304) of the users based on the overlapping features.3.The method as claimed in claim 1 or claim 2, wherein obtaining the connectivity score vector comprises:obtaining a first pruned group knowledge graph by pruning the group knowledge graph (306) based on user-user relationships; andobtaining the connectivity score vector by processing the first pruned group knowledge graph using a first graph-based neural network.4.The method as claimed in any one of claims 1 to 3, wherein obtaining the group profile comprises:obtaining, for each user in the group, exposable interest embedding based on the group knowledge graph (306), wherein the exposable interest embedding indicates the respective user's content preferences exposable to the group;modifying the group knowledge graph (306) based on the exposable interest embedding to obtain a final group knowledge graph; andobtaining the group profile based on the group connectivity score vector and the final group knowledge graph.5.The method as claimed in any one of claims 1 to 4, further comprising:obtaining a second pruned knowledge graph by pruning the group knowledge graph (306) based on user-content relationships;evaluating, using a second graph-based neural network, content preferences of each user relevant to exposure to the group based on the second pruned knowledge graph; andobtaining the exposable interest embedding based on the evaluation.6.The method as claimed in any one of claims 1 to 5, wherein recommending content to the group comprises:encoding the group profile to obtain a group profile embedding;obtaining a plurality of contents for recommendation;determining a rank for each of the plurality of contents based on the group profile embedding; andrecommending content to the group based on the rank for each of the plurality of contents.7.The method as claimed in claim 6, wherein recommending content to the group based on the rank for each of the plurality of contents comprises:obtaining multimedia data from the plurality of contents;obtaining a multimedia preferences based on the group profile;obtaining candidate multimedia assets based on the multimedia preferences and the multimedia data;computing a contrastive loss between the candidate multimedia assets and exposable interest embedding indicating the respective user's content preferences exposable to the group;selecting at least one multimedia asset from the candidate multimedia assets for at least a subset of the plurality of contents based on the contrastive loss; andrecommending content to the group based on the at least one multimedia asset and the rank for each of the plurality of contents.8.The method as claimed in any one of claims 1 to 7, further comprising:updating the knowledge graph of each user based on response from the group.9.An electronic device (200) for generating content recommendations for a group of users, the electronic device (200) comprising:memory (202) storing instructions; andone or more processing unit (204) operatively coupled to the memory (202) and comprising processing circuitry, wherein the one or more processing unit (204) individually or collectively executes the instructions to cause the electronic device (200) to:obtain a group knowledge graph (306) for a group based on a knowledge graph (304) of each user in the group, wherein the knowledge graph (304) of each user is associated with respective user activity performed on one or more devices;obtain, for each user in the group, a connectivity score vector based on the group knowledge graph (306), wherein the connectivity score indicates contextual contribution of the user in content recommendations with respect to other users in the group;obtain a group connectivity score vector based on the connectivity score vectors of each user in the group;obtain, based on the group connectivity score vector and the group knowledge graph (306), a group profile representing collective content preference of the group; andrecommend content to the group based on the group profile.10.The electronic device (200) as claimed in claim 9, wherein, the one or more processing unit (204) individually or collectively executes the instructions to cause the electronic device (200) to:detect the group for which content is to be recommended;obtain the knowledge graph (304) of each user in the group from a database including knowledge graphs of a plurality of users;identify overlapping features among the knowledge graphs (304) of the users in the group; andobtain the group knowledge graph (306) by integrating the knowledge graphs (304) of the users based on the overlapping features.11.The electronic device (200) as claimed in claim 9 or claim 10, wherein the one or more processing unit (204) individually or collectively executes the instructions to cause the electronic device (200) to:obtain a first pruned group knowledge graph by pruning the group knowledge graph (306) based on user-user relationships; andobtain the connectivity score vector by processing the first pruned group knowledge graph using a first graph-based neural network.12.The electronic device (200) as claimed in any one of claims 9 to 11, wherein the one or more processing unit (204) individually or collectively executes the instructions to cause the electronic device (200) to:obtain, for each user in the group, exposable interest embedding based on the group knowledge graph (306), wherein the exposable interest embedding indicates the respective user's content preferences exposable to the group;modify the group knowledge graph (306) based on the exposable interest embedding to obtain a final group knowledge graph; andobtain the group profile, based on the group connectivity score vector and the final group knowledge graph.13.The electronic device (200) as claimed in any one of claims 9 to 12, wherein the one or more processing unit (204) individually or collectively executes the instructions to cause the electronic device (200) to:obtain a second pruned knowledge graph by pruning the group knowledge graph (306) based on user-content relationships;evaluate, using a second graph based neural network, content preferences of each user relevant to exposure to the group, based on the second pruned knowledge graph; andobtain the exposable interest embedding based on the evaluation.14.The electronic device (200) as claimed in any one of claims 9 to 13, wherein the one or more processing unit (204) individually or collectively executes the instructions to cause the electronic device (200) to:encode the group profile to obtain a group profile embedding;obtain a plurality of contents for recommendation;determine a rank for each of the plurality of contents based on the group profile embedding; andrecommend content to the group based on the rank for each of the plurality of contents.15.A computer-readable medium including instructions, which, when executed by at least one processing unit (204) individually or collectively, cause the electronic device (200) to perform a method of any one of claims 1 to 8.