Composite Graph Prediction for Mobile App Installation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for predicting user behavior in networks face challenges due to individual variance in app installation, where some users install apps without external influence, and exogenous factors like app popularity are not adequately considered, making it difficult to model the spreading of apps effectively.
Innovation Solution
The solution involves creating a composite graph from multiple network types (call log, Bluetooth proximity, friendship, and affiliation) weighted by an optimization vector, and incorporating exogenous factors through a virtual graph to predict app installation probabilities, allowing for the selection of users for targeted promotions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple network graphs are combined to predict user behavior, then prediction accuracy improves, but system complexity increases
Solution Approach 1:
The system segments the complex prediction problem into multiple independent network graphs (call log graph, text message graph, Bluetooth proximity graph, co-location graph, friendship graph, affiliation graph), each representing a specific type of user interaction. Each graph is processed separately with its own weighting, and the results are aggregated to form the final prediction, making the overall system more manageable and interpretable
Solution Approach 2:
The system creates a composite graph by combining multiple different network graphs with different weighting factors. This composite structure integrates information from diverse sources (communication patterns, physical proximity, social relationships) to form a more comprehensive and accurate representation of user behavior than any single graph could provide alone
2Measurement precision
If exogenous factors are incorporated into the prediction model, then prediction accuracy improves, but computational requirements increase
Solution Approach 1:
The system introduces a virtual graph as an intermediary structure to represent exogenous factors such as app quality and overall popularity. This virtual graph acts as a mediator that captures external influences without requiring direct computation of complex interactions between all users and all exogenous factors, reducing computational overhead while maintaining prediction accuracy
3Productivity
If targeted user selection is implemented, then resource allocation efficiency improves, but selection precision requirements increase
Solution Approach 1:
The system changes the parameter of user selection from binary (selected/not selected) to probabilistic (conditional probability of installation). By computing conditional probabilities for each user based on their position in the composite graph and the diffusion model, the system can rank users by their likelihood of adoption, enabling efficient resource allocation to the most promising users without requiring absolute certainty in selection
Data Source
AI summary
In exemplary implementations of this invention, mobile application (app) installations by users of one or more networks are predicted. Using network data gathered by smartphones, multiple “candidate” graphs (including a call log graph) are calculated. The “candidate” graphs are weighted by an optimization vector and then summed to calculate a composite graph. The composite graph is used to predict the conditional probabilities that the respective users will install an app, depending in part on whether the user's neighbors have previously installed the app. Exogenous factors, such as the app's quality, may be taken into account by creating a virtual candidate graph. The conditional probabilities may be used to select a subset of the users. Signals may be sent to the subset of users, including to recommend an app. Also, the probability of successful “trend ignition” may be predicted from network data.