Deep Network Resource Classification for Spam Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Internet search engines face challenges in accurately identifying and categorizing resources, particularly in distinguishing between relevant and spam content, which affects the quality of search results provided to users.
Innovation Solution
A deep network system is employed to classify resources by processing features through embedding functions and neural network layers, generating category scores that predict the likelihood of a resource belonging to specific categories, such as spam or resource types, to improve search engine efficiency and user satisfaction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional classification methods are used to categorize resources, then the system is simpler to implement, but the accuracy of distinguishing spam from relevant content deteriorates
Solution Approach 1:
The patent segments the classification task into multiple independent feature extraction components, each handling specific aspects of resource analysis (text content, metadata, structural properties). These segmented features are then processed separately before being combined for final classification, improving accuracy while managing complexity through modular organization
Solution Approach 2:
The patent transforms the classification problem from traditional single-dimension analysis to multi-dimensional feature space analysis by extracting numerous diverse features (textual, structural, metadata) and processing them through multiple neural network layers, enabling more accurate discrimination of spam resources
2Measurement precision
If comprehensive feature analysis is performed on all resources, then classification accuracy improves, but processing time increases
Solution Approach 1:
The patent performs preliminary feature extraction and analysis on all incoming resources before the actual classification decision is made. By pre-computing feature vectors and organizing them in structured formats, the system prepares data in advance, reducing the time required for final classification while maintaining comprehensive analysis
Solution Approach 2:
The patent implements a two-stage processing approach where not all resources receive the full comprehensive analysis. Quick-filter rules and preliminary scoring identify obviously spam or clearly legitimate resources that can be classified with less processing, while only borderline cases undergo complete multi-dimensional feature analysis
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 using one or more neural network layers to generate an alternative representation of the features, 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 using a classifier to generate a respective category score for each category in a pre-determined set of categories, wherein each of the respective category scores measure a predicted likelihood that the resource belongs to the corresponding category.