Machine Learning Ranking Model for Search Relevance and Conversion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing predictive data analysis solutions have low conversion rates due to reliance on search algorithms that output relevant but not user-oriented search results, and content items are not optimized for cost, quality of service, and ratings.
Innovation Solution
The method involves generating label sets based on user search session data and transaction data, determining a dominant label set, and using it to assign labels to search query-content item record pairs in a training dataset, thereby updating machine learning model parameters to improve content item suggestion relevance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional search algorithms are used to output search results, then relevance to search query is improved, but conversion rate (user transaction) deteriorates
Solution Approach 1:
The patent merges traditional search relevance signals with user-specific activity data and contextual information into a unified ranking model. This combination allows the system to maintain query relevance while incorporating user behavior patterns that drive conversions, resolving the contradiction between search result relevance and conversion rate.
Solution Approach 2:
The system dynamically adjusts search result rankings based on real-time user activity data, session context, and individual user profiles. This dynamic adaptation enables the same search algorithm to optimize for both relevance and conversion by continuously learning from user interactions and adjusting predictions accordingly.
2Measurement precision
If content items are selected based on traditional relevance metrics, then accuracy of search results is improved, but optimization for cost and quality of service deteriorates
Solution Approach 1:
The patent introduces new ranking parameters beyond traditional relevance metrics, incorporating user-specific weights for cost, quality of service, and ratings. By changing the optimization parameters dynamically based on user preferences and contextual factors, the system maintains accuracy while adapting to different optimization goals.
Solution Approach 2:
The system applies different quality criteria and weighting factors to different content items and users locally. Instead of a uniform relevance metric, the system tailors the evaluation criteria to match individual user preferences and contextual requirements, enabling simultaneous optimization for accuracy, cost, and service quality.
3Productivity
If personalized user activity data is integrated into search algorithms, then conversion rate is improved, but system complexity deteriorates
Solution Approach 1:
The patent segments the complex personalization task into independent modular components: user profile module, activity data processing module, ranking optimization module, and feedback learning module. This segmentation reduces overall system complexity by allowing each component to be developed, optimized, and maintained independently while working together to improve conversion rates.
Data Source
AI summary
Various embodiments of the present disclosure provide methods, apparatus, systems, computing devices, computing entities, and/or the like for generating (i) a first label set representative of a selection of a content item based on search session data and (ii) a second label set representative of one or more transactions associated with the content item based on transaction data. A dominant label set is determined from the plurality of label sets based on an occurrence frequency associated with the first label set and the second label set. Based on an occurrence of an event associated with the dominant label set, either a first label associated with the dominant label set is assigned to first search query-content item record pairs associated with a training dataset, or one or more stochastic labels from the plurality of label sets are assigned to second search query-content item record pairs associated with the training dataset.