Application-Centric Predictive Analytics Broker for Interactive Apps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional predictive analytics approaches, particularly analyst-centric methods, create a disconnect between analysts and developers in interactive applications, where analysts develop models based on available data but the results are often unusable by the interactive application, and developers struggle to implement these results effectively.
Innovation Solution
An application-centric approach is introduced, where a broker device receives requests for action data and session data from interactive applications, determines scoring package data using predictive models, and generates action data to be transmitted back to the application, facilitating a seamless integration of predictive analytics within the application's decision-making process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If an analyst-centric approach is used to develop predictive models, then the model can be created based on available data, but the results are often unusable by the interactive application and developers struggle to implement them effectively
Solution Approach 1:
The patent introduces a scoring framework as an intermediary layer between the predictive model and the interactive application. This framework includes scoring packages that contain models, metrics, and thresholds, acting as a mediator that translates complex analytical results into actionable, easily implementable scoring outputs that developers can readily use without directly handling the complexity of the underlying predictive models
Solution Approach 2:
The patent transforms the output parameters of predictive models into standardized scoring parameters with defined thresholds and metrics. By changing the parameter representation from raw model outputs to structured scoring values, the system makes the results more usable and easier to implement in interactive applications while maintaining the accuracy of the underlying predictive analysis
2Loss of information
If the analyst develops a model with detailed data requirements, then the predictive analysis can be comprehensive, but the developer cannot accommodate the analyst's data requests in real-time
Solution Approach 1:
The patent implements preliminary action by pre-configuring scoring packages with all necessary models, metrics, and data requirements before runtime. The scoring framework is prepared in advance with predefined thresholds and evaluation criteria, so that during real-time interactive application execution, the system can directly apply these pre-prepared scoring mechanisms without needing to gather or process additional data on demand
Solution Approach 2:
The patent creates a dynamic scoring framework that can adapt to different data availability scenarios. The scoring packages are designed to be flexible and can operate with varying levels of data input, allowing the system to maintain comprehensive predictive analysis capabilities while adapting to the real-time data constraints of interactive applications
3Measurement precision
If the interactive application requests detailed session data for predictive analysis, then the predictive recommendations can be more accurate, but the application complexity and data processing requirements increase
Solution Approach 1:
The patent segments the predictive analytics system into distinct modular components: scoring packages, metrics, thresholds, and evaluation rules. Each scoring package is an independent, self-contained unit that can be developed, tested, and deployed separately. This segmentation reduces application complexity by breaking down the monolithic predictive analysis into manageable, interchangeable modules that can be selectively applied based on specific application needs
Data Source
AI summary
A device, system, and method use predictive analytics based on an application-centric approach. The method includes receiving a request from an interactive application interacting with a user utilizing a user device for action data indicating an action to be taken by the interactive application during a session with the user device, the request generated by the interactive application based on a decision point associated with the interactive application. The method includes receiving from the interactive application session data associated with the session and the user device. The method includes determining scoring package data associated with the request based on the session data, the scoring package data comprising a predictive model indicative of a plurality of actions to be performed by the interactive application. The method includes generating the action data based on the scoring package data and transmitting the action data to the interactive application.