Deep Network Scoring Concept Terms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods fail to accurately identify and select relevant concept terms for resources, particularly in online advertisement auctions, leading to less relevant advertisements being displayed.
Innovation Solution
A deep network system that processes resource features using embedding functions and neural networks to generate relevance scores for concept terms, allowing for the selection of relevant advertising keywords by filtering out terms that do not match the resource's language or have high TF-IDF scores.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional methods are used to identify concept terms for resources, then the system is simpler and easier to implement, but the accuracy of identifying relevant concept terms deteriorates
Solution Approach 1:
The patent introduces embedding functions as intermediary components that transform resource features into a unified vector space. These embedding functions act as mediators between the raw features and the concept term scoring mechanism, enabling accurate relevance assessment without requiring direct complex interactions between all feature pairs.
Solution Approach 2:
The system segments the concept term identification task into distinct processing stages: feature extraction, embedding transformation, and relevance scoring. Each stage handles specific aspects of the problem independently, improving overall accuracy while managing complexity through modular organization of the deep network components.
2Measurement precision
If a deep network with embedding functions is used to process features, then the accuracy of predicting relevant concept terms improves, but the computational complexity increases
Solution Approach 1:
The embedding functions perform preliminary transformation of resource features into standardized vector representations before the main scoring process. This pre-processing step consolidates feature information in advance, reducing the computational burden during the actual concept term relevance prediction phase.
3Quantity of substance
If more concept terms are selected for advertisement auctions, then the coverage of relevant advertisements increases, but the relevance quality of selected advertisements deteriorates
Solution Approach 1:
The system uses relevance scores generated by the deep network as feedback to rank and select concept terms. This feedback mechanism ensures that only concept terms with sufficiently high relevance scores are selected for advertisement auctions, maintaining quality while achieving appropriate coverage through score-threshold-based filtering.
Data Source
AI summary
Methods, systems, and apparatus, including computer programs encoded on computer storage media, for scoring concept terms using a deep network. One of the methods includes receiving an input comprising a plurality of features of a resource, wherein each feature is a value of a respective attribute of the resource; processing each of the features using a respective embedding function to generate one or more numeric values; processing the numeric values to generate an alternative representation of the features of the resource, wherein processing the floating point values comprises applying one or more non-linear transformations to the floating point values; and processing the alternative representation of the input to generate a respective relevance score for each concept term in a pre-determined set of concept terms, wherein each of the respective relevance scores measures a predicted relevance of the corresponding concept term to the resource.