ML Floor Bid Updating for Dynamic Real-Time Auctions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for setting and adjusting floor bid values in real-time computer-implemented auctions are inefficient and fail to adapt to dynamic market conditions, leading to suboptimal outcomes and inefficiencies.
Innovation Solution
Utilizing machine learning models, specifically an XGBoost model, to dynamically optimize floor bid values by analyzing historical auction bid data and retraining periodically to align with current demand, adjusting values based on real-time market changes and user group characteristics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional methods are used to set and adjust floor bid values, then the system is simple to operate, but the system fails to adapt to dynamic market conditions leading to suboptimal auction outcomes
Solution Approach 1:
The patent implements dynamic floor bid values that automatically adjust based on real-time market conditions, user group characteristics, and historical auction data. The system transitions from static, manually-set floor values to dynamically optimized values that adapt continuously through machine learning models, resolving the contradiction between adaptability and complexity by automating the adjustment process.
Solution Approach 2:
The system employs self-service mechanisms where machine learning models automatically analyze market conditions and determine optimal floor bid values without requiring manual intervention. The models periodically retrain themselves using new auction data, enabling the system to adapt to changing market conditions autonomously while maintaining simplicity for operators.
2Productivity
If machine learning models are used to dynamically optimize floor bid values, then auction efficiency and effectiveness are improved, but computational complexity and processing time increase
Solution Approach 1:
The system performs preliminary actions by pre-training machine learning models using historical auction data before actual auctions occur. The models are periodically retrained in advance with accumulated data, so that during real-time auctions, the pre-trained models can quickly provide optimized floor bid values without adding significant processing delays to the auction workflow.
Solution Approach 2:
The system implements periodic action by retraining machine learning models at scheduled intervals rather than continuously. This approach balances the need for up-to-date market adaptation with computational efficiency, allowing the system to maintain high auction processing speeds while periodically updating the models with new data to improve their predictive accuracy.
3Productivity
If floor bid values are adjusted frequently to match market demand, then auction outcomes are maximized, but system stability and reliability decrease
Solution Approach 1:
The system employs feedback mechanisms where machine learning models continuously monitor auction outcomes, market conditions, and bidding patterns. The models use this feedback to make gradual, data-driven adjustments to floor bid values that optimize auction outcomes while maintaining stability. The feedback loop ensures that changes are based on actual market responses rather than speculative adjustments.
Solution Approach 2:
The system carefully manages parameter changes by adjusting floor bid values in controlled increments based on machine learning model predictions. Rather than making large, frequent changes that could destabilize the system, the models analyze multiple data points and make measured parameter adjustments that optimize outcomes while preserving system reliability and predictability.
Data Source
AI summary
A computer-implemented method comprising collecting bid data corresponding to bids of different user computers in a real-time computer-implemented auction, wherein each user computer is associated with a digital content item that a content server can serve to a web server; based on bid data parameters of the bid data, aggregating digital identifiers of user computers into a plurality of user groups, and digitally storing a training dataset comprising the bid data parameters and the user groups; training a machine learning model (MLM) using the training dataset to create and store a trained MLM; executing an inference stage of the trained MLM over input data comprising characteristics of a particular user group from among the plurality of user groups and revenue-per-thousand impressions data (RPM data) to output a prediction of a group-associated floor bid value for the particular user group; transmitting, to the content server, the floor bid value and an identifier of the particular user group; and dynamically adjusting, by the content server, a frequency of updating the floor bid value, wherein the adjusting occurs when an average maximum bid price for a set of auctions exceeds the floor bid value.