Query Prediction Model Using Time Rank Category Functions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing search systems struggle to accurately predict user interests, especially when users submit queries infrequently or in different contexts, leading to irrelevant advertisements and search results.
Innovation Solution
A machine-learned prediction model is trained using query log data and category data, incorporating time-based, rank-based, and category-based prediction functions to generate category predictions for next queries, balancing query frequency and submission times to improve prediction reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If query history is used to predict user interests, then prediction coverage is improved, but prediction accuracy deteriorates when users submit queries infrequently or in different contexts
Solution Approach 1:
The patent applies dynamics by making the prediction model adaptive and context-aware. Instead of using a static query history approach, the system dynamically adjusts predictions based on user context, device type, and query patterns. The model evolves its predictions based on real-time user behavior and environmental factors, resolving the contradiction between broad coverage and accurate predictions for infrequent users.
Solution Approach 2:
The patent changes parameters by incorporating multiple features beyond simple query history, including user context, device information, and temporal patterns. By adjusting the weighting and consideration of these parameters based on user behavior patterns, the system achieves both broad prediction coverage and maintained accuracy even for users who query infrequently or from different contexts.
2Reliability
If multiple prediction functions are incorporated, then prediction reliability is improved, but model complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the prediction model into distinct modular functions: time-based prediction functions, rank-based prediction functions, and category-based prediction functions. Each function handles specific aspects of user behavior analysis independently, improving reliability through diversified prediction approaches while managing complexity through modular design that allows selective deployment and independent optimization of each component.
Data Source
AI summary
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for predicting user interests. In one aspect, a method includes training a prediction model to generate a category prediction of a next query from a set of queries, the category prediction specifying categories to which a next query belongs, the prediction model comprising a plurality of prediction functions that each generate a category prediction of a next query, wherein the prediction functions include two or more of a time-based prediction functions that generate a category prediction based on the category data and a difference of submission times of the queries, a rank-based prediction functions that generate a category prediction based on the category data and a rank order by which the queries were received, and a category-based prediction function that generates a category prediction based on the category data.