Classifying resources using a deep network
Topics: AI (Deep Learning), Document Classification, E-E-A-T, Indexing, Probably in use, Ranking
The Google patent describes a system for classifying web resources into categories using a deep neural network. It includes an embedding layer that converts input features into numeric values, neural network layers that apply non-linear transformations to generate an alternative representation, and a classifier that outputs category scores. The system can be used to detect spam websites or classify resources into different types. The scores can be used by search engines to determine indexing, ranking, and presentation of search results.