Account Recommendation System Using Inverse Follower Scoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional social networking systems prioritize recommending popular accounts with a large number of followers, leading to a cyclical effect where only popular accounts gain more followers, while less popular accounts are discouraged, resulting in an environment where only high-followers are encouraged to post content, and others are discouraged due to lack of growth.
Innovation Solution
Accounts are grouped based on the number of followers and assigned value scores inversely related to their average number of followers, with a machine learning model trained to uprank less popular accounts and downrank highly popular ones, using a learning to rank algorithm like LambdaMART to select recommendations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If the system recommends popular accounts with large number of followers, then the recommendation simplicity and user familiarity are improved, but the diversity of recommended accounts and growth opportunity for less popular accounts deteriorates
Solution Approach 1:
The patent segments accounts into different popularity tiers (e.g., high-followers, medium-followers, low-followers) and applies different recommendation strategies to each segment. This allows the system to maintain simple recommendations for popular accounts while also promoting diverse less popular accounts, resolving the contradiction between recommendation simplicity and account diversity.
Solution Approach 2:
The system changes the recommendation parameter from purely follower-count-based to a multi-factor parameter that includes follower count, account growth potential, user interest matching, and diversity factors. This parameter transformation enables the system to balance between recommending established popular accounts and discovering diverse emerging accounts.
2Productivity
If the system prioritizes popular accounts for recommendations, then the user engagement with well-known accounts is improved, but the follower growth rate and activity of less popular accounts deteriorates
Solution Approach 1:
The system implements a feedback mechanism where recommendation outcomes are continuously monitored. When less popular accounts receive recommendations and gain followers, this positive feedback is used to adjust future recommendations, creating a virtuous cycle that promotes both engagement and equitable follower distribution across different account popularity levels.
Solution Approach 2:
The system performs preliminary actions by proactively recommending less popular accounts to users before these accounts would naturally gain visibility. This early intervention helps less popular accounts build their follower base and engagement metrics, enabling them to compete more effectively with popular accounts in future recommendation cycles.
3Power
If the system uses traditional recommendation algorithms, then the computational efficiency is improved, but the ability to promote equitable growth across different account popularity levels deteriorates
Solution Approach 1:
The patent segments the recommendation process into multiple stages: initial filtering by basic criteria, scoring based on multiple factors including growth potential, and final selection with diversity constraints. This segmented approach maintains computational efficiency at each stage while achieving the complex goal of promoting equitable growth across different account popularity levels.
Solution Approach 2:
The system transforms the recommendation algorithm from a simple sorting operation to a multi-objective optimization problem with parameters for engagement probability, growth potential, and diversity. This parameter transformation enables the algorithm to balance computational efficiency with the ability to promote equitable growth across different account types.
Data Source
AI summary
Systems, methods, and non-transitory computer-readable media can identify a set of accounts, each account of the set of accounts having a number of followers. The set of accounts are grouped into a plurality of groups based on number of followers, wherein each group is associated with a value score. A machine learning model is trained using a set of training data comprising account recommendation conversion information, wherein the account recommendation conversion information comprises a plurality of successful account recommendations, and each successful account recommendation is assigned a weight based on the value scores associated with the plurality of groups. One or more accounts of the set of accounts are selected to present as account recommendations based on the machine learning model.