Social Influence Weighting in Information Recommendation Systems

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current information recommendation methods in social applications rely heavily on user interest levels, which may not accurately predict interaction probabilities, leading to suboptimal effectiveness in pushing recommended information.

Innovation Solution

The method determines the influence degree of friends who have interacted with shared information on the likelihood of a user interacting with recommended information, integrating this data to enhance the accuracy of interaction probability predictions and improve information pushing effectiveness.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If user interest level based on relevance degree is used to determine information pushing, then the system is simple to operate, but the accuracy of predicting user interaction is insufficient

Engineering Contradiction:
Improveaccuracy of predicting user interactionVSAvoidcomplexity of recommendation system
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent combines multiple factors including user interest level, friend interaction data, and social relationship strength into a unified recommendation model. By merging these different dimensions of data, the system achieves more accurate interaction probability prediction while maintaining operational simplicity through automated integration of multiple variables.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent introduces friend interaction data as an intermediary factor that mediates between user interest and information recommendation. This intermediary element provides additional contextual information about user behavior patterns, enabling more accurate prediction without requiring direct complex analysis of user preferences alone.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Productivity

If only user interest level is considered for information recommendation, then the system complexity is low, but the effectiveness of information pushing is suboptimal

Engineering Contradiction:
Improveeffectiveness of information pushingVSAvoidcomplexity of data processing
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent performs preliminary analysis of friend interaction data and social relationship patterns before generating recommendations. By pre-processing and storing interaction histories and relationship strengths, the system can quickly retrieve and utilize this data during recommendation generation, improving effectiveness without adding significant processing complexity at decision time.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent incorporates feedback from friend interactions with recommended information into the recommendation model. By monitoring how friends interact with shared content and using this feedback to adjust recommendation strategies, the system continuously improves information pushing effectiveness while maintaining a relatively simple operational framework.

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP3404556B1Information recommendation method and apparatus, and server
Publication Date: 2023.07.12 TENCENT TECHNOLOGY (SHENZHEN) CO LTD
  • EP3404556B1 patent drawingFigure 1
  • EP3404556B1 patent drawingFigure 2
  • EP3404556B1 patent drawingFigure 3

AI summary

An information recommendation method and apparatus, and a server. The method comprises determining, from friends of a target user, a target friend that has interacted with target recommended information (S100); determining interaction data, regarding previously shared information published by the target friend, of the target user (S110); according to the determined interaction data, regarding the previously shared information published by each target friend, of the target user, determining the influence degree of the target friend on the interaction between the target user and the target recommended information (S120); according to the influence degree of the target friend on the interaction between the target user and the target recommended information, determining a target influence degree (S130); according to the target influence degree, determining the possibility degree of the interaction between the target user and the target recommended information (S140); and if the possibility degree meets a pre-set condition, pushing the target recommended information to the target user (S150). The method, the apparatus and the server improve the accuracy of the determined possibility of interaction between a user and recommended information, and improves the validity of the pushing of the recommended information.