Browsing History Search Using Semantic Resource Embeddings
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Browser histories lack sufficient searchable text for effective natural language searches, and memory constraints prevent storing and indexing the content of visited resources, making it difficult to recall previously viewed websites.
Innovation Solution
Implement a memory-efficient semantic representation using generative models to generate embeddings for resource content, allowing retrieval of previously visited resources based on content similarity within memory constraints.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the content of visited resources is stored and indexed for natural language search, then search accuracy is improved, but memory usage exceeds device constraints
Solution Approach 1:
The patent extracts only the essential semantic meaning from webpage content by generating embeddings, rather than storing the full content. This extraction process converts large amounts of text into compact vector representations that capture the essential information needed for search while occupying minimal memory space.
Solution Approach 2:
The patent transforms the content from its original text form into a different parameter representation (embeddings/vectors). This parameter change allows the system to represent complex content in a condensed numerical format that is both memory-efficient and suitable for similarity-based search operations.
2Quantity of substance
If only URL and title are stored in browser history, then memory usage is reduced, but searchability of history visits deteriorates
Solution Approach 1:
The patent creates a compressed copy of the content in the form of embeddings. Instead of storing the actual content text, the system stores these embedded representations that can be used for search purposes, effectively copying only the search-relevant information in a space-efficient manner.
Data Source
AI summary
The present disclosure provides a system and methods for providing responses to user provided questions that are grounded in the user's browsing history and enables searching of the history for resources previously viewed based on the content of the resources. In one example embodiment, a portion of a query is received. A query suggestion relevant to the portion of the query is obtained. A resource from a history repository relevant to the portion of the query is identified by obtaining a semantic representation of the portion of the query, and identifying a semantic representation of content associated with the resource from the history repository. The semantic representation of the content includes a similarity score with the semantic representation of the query that satisfies a threshold. A resource suggestion is generated for the resource. The query suggestion and the resource suggestion is provided as selectable query completions.