Mobile App Keyword Extraction via Similar Application Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Advertisers face challenges in presenting sponsored content items to users who are likely to be interested, as existing methods rely on keywords associated with applications, which may not accurately reflect user interests due to limited keyword data, especially across different mobile application platforms.
Innovation Solution
The method involves obtaining information about a mobile application of interest, determining similar applications, extracting new keywords using a model trained on statistical information from these similar applications, and expanding keywords based on statistical data from counterpart applications on other platforms, ensuring more relevant sponsored content is presented.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If keyword data is limited to what is directly associated with an application, then the system maintains simplicity in data collection, but the accuracy of user interest representation deteriorates
Solution Approach 1:
The system performs preliminary keyword extraction from similar applications before the actual sponsored content presentation. By pre-extracting keywords from multiple similar applications and building a comprehensive keyword set in advance, the system improves user interest representation without adding complexity during the actual content delivery phase. The keyword extraction and statistical information collection is done beforehand, so when sponsored content needs to be presented, the system can directly use the pre-prepared keyword set.
2Quantity of substance
If the system collects statistical information from multiple similar applications across different platforms, then the quantity of keyword data increases, but the difficulty of detecting and measuring accurate user interests increases
Solution Approach 1:
The system implements feedback mechanisms by collecting statistical information about keyword performance from multiple similar applications. This statistical information serves as feedback that helps refine and adjust the keyword set for sponsored content presentation. By analyzing which keywords perform well across similar applications, the system can iteratively improve its user interest detection accuracy, transforming the complexity of multiple data sources into a structured feedback loop that enhances measurement capability.
3Reliability
If the system uses a model trained on statistical information from similar applications, then the reliability of keyword extraction improves, but the device complexity increases
Solution Approach 1:
The system uses copying by creating a statistical model based on data from similar applications. Instead of directly analyzing each application's user behavior in real-time, the system copies the statistical patterns and keyword performance metrics from similar applications into a trained model. This model can then be applied to extract keywords for sponsored content with high reliability without requiring complex real-time analysis infrastructure, as the heavy computational work has already been done during model training.
Data Source
AI summary
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for content presentation. In one aspect, a method includes obtaining information associated with a mobile application of interest; determining a plurality of similar applications to the application of interest; determining keywords from the similar applications; and extracting new keywords for the application of interest using a model trained using statistical information for keywords of the plurality of similar applications that overlap with keywords of the application of interest.