Inference Detection via Sponsored Ad Correlation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing information retrieval systems fail to accurately detect inferences between terms due to reliance on independent attributes, leading to inconsistent correlation between objects, which affects applications like Internet search engines and document categorization.
Innovation Solution
A system that performs inference detection by analyzing correlations between topic words and sponsored advertisements, using bipartite graphs and similarity metrics to cluster topic words and URLs based on shared advertisement links, thereby identifying related terms and products without requiring a priori knowledge of retailer brands or private data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing systems use hyperlinks or keyword overlap to measure inference between terms, then the systems can detect inference relationships, but the correlation accuracy between objects deteriorates due to reliance on independent attributes created for different purposes
Solution Approach 1:
The patent introduces sponsored advertisements as an intermediary element that connects topic words. Instead of directly comparing topic words using independent attributes, the system uses ads as a mediator that naturally associates multiple topic words through commercial intent, thereby improving correlation accuracy and consistency
Solution Approach 2:
The system changes the measurement parameter from traditional hyperlink/keyword overlap metrics to sponsored advertisement co-occurrence metrics. This parameter change leverages the inherent semantic relationships captured in advertising data, which are created for commercial purposes rather than information retrieval, thus improving inference detection accuracy
2Ease of manufacture
If the system uses sponsored advertisement data for inference detection, then the need for extensive database maintenance is reduced, but the system complexity increases due to bipartite graph generation and collapsing operations
Solution Approach 1:
The system leverages publicly available sponsored advertisement data that is already structured and maintained by advertising platforms. This self-service approach allows the system to access ready-to-use data without requiring extensive manual database maintenance, as the advertising ecosystem automatically generates and updates the data
Solution Approach 2:
The patent segments the problem into distinct graph processing stages: generating a bipartite graph from advertisement data, collapsing the bipartite graph to extract topic word relationships, and clustering results. This segmentation makes the complex processing manageable and modular, reducing overall system complexity
Data Source
AI summary
One embodiment of the present invention provides a system that performs inference detection based on Internet advertisements. In doing so, this system first receives a set of topic words, performs a search query on each topic word using a search engine, and gathers a set of Uniform Resource Locators (URLs) associated with sponsored advertisement from the search results corresponding to each search query. Then, the system determines a correlation between two topic words based on their corresponding URLs associated with sponsored advertisement, and produces a result which indicates groups of correlated topic words.