Deep Content Ranking Using Interaction Probabilities for Short Queries
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional retrieval systems struggle with accurately ranking content items when query inputs are limited to a few characters or contain typographical errors, leading to irrelevant results and increased resource consumption.
Innovation Solution
A deep machine learning model is employed to generate ranking scores for content items based on query and content features, predicting probabilities of user-item interactions to improve relevance ranking, even with minimal query input.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional retrieval systems rely heavily on query text for ranking, then ranking accuracy improves for complete queries, but performance deteriorates when query input is limited to a few characters or contains typographical errors
Solution Approach 1:
The patent introduces an intermediary deep machine learning model that sits between the query input and the ranking decision. This ML model acts as a mediator that processes limited or erroneous query inputs and transforms them into meaningful ranking scores by learning from historical user behavior patterns, thereby resolving the contradiction between relying on query text and handling minimal/erroneous inputs
Solution Approach 2:
The system changes the parameters used for ranking from purely query-text-based metrics to a combination of query features and user behavior features. By incorporating parameters such as user interaction history, content popularity, and engagement patterns, the system can maintain high ranking accuracy even when query input is minimal or contains errors
2Speed
If retrieval systems retrieve items based on minimal query input, then response speed improves, but the number of irrelevant items increases
Solution Approach 1:
The patent replaces conventional mechanical text-matching mechanisms with a deep machine learning-based ranking system. This substitution allows the system to quickly process minimal query inputs and immediately provide relevant results by leveraging learned patterns from user behavior data, thereby reducing the number of irrelevant items while maintaining fast response speed
3Measurement precision
If ranking algorithms rely on query text, then relevance ranking improves for complete queries, but computing resources are consumed when users submit multiple queries to correct errors
Solution Approach 1:
The system incorporates feedback mechanisms where users can indicate whether retrieved items are relevant or not. This feedback is fed back into the machine learning model to continuously improve its ranking accuracy. By doing so, the system can more accurately predict user preferences from minimal inputs, reducing the need for multiple corrective queries and thereby lowering computing resource consumption while maintaining high relevance ranking
Data Source
AI summary
Disclosed herein are system, method and/or computer program product embodiments, and/or combinations and sub-combinations thereof, for ranking a plurality of content items for presentation to a user. An embodiment generates a ranking score for each content item by: providing input to a deep machine learning (ML) model, the input including at least one or more query features and one or more content item features, determining, by the deep ML model and based at least on the input, a first probability of a first type of interaction between the user and the content item and a second probability of a second type of interaction between the user and the content item, and calculating the ranking score for the content item based at least on the first and second probabilities. An embodiment ranks the content items for presentation based on the ranking score associated with each content item.