Named Entity Extraction for Web Search Relevance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The vast and growing amount of information on the internet makes it difficult for users to find relevant and focused resources, as conventional search engines struggle to accurately return named entities and related information, leading to challenges in refining queries and improving page relevance.
Innovation Solution
A system that includes an extraction component for identifying named entities in documents, an indexing component for efficient storage, and a retrieval component that computes entity scores and thresholds to return relevant entities to users, enhancing the relevance of search results by aggregating entity information from top matching documents.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional search engines return links and query suggestions, then the search results are comprehensive, but the relevance and focus of the results deteriorate
Solution Approach 1:
The patent extracts named entities from the vast amount of web information and presents them as focused search results. The extraction component identifies and isolates specific entities (people, organizations, locations, products) from documents, allowing users to obtain concise, high-relevance results rather than being overwhelmed by comprehensive but diffuse information.
Solution Approach 2:
The search system segments information by entity type and relevance, organizing the vast information landscape into discrete, manageable categories. The extraction component identifies specific entity mentions, and the retrieval component returns segmented results grouped by entity type, making the information more digestible and relevant.
2Adaptability or versatility
If the amount of information and data sources continue to grow, then the coverage and comprehensiveness improve, but the difficulty of finding relevant resources increases
Solution Approach 1:
The patent introduces named entity extraction as an intermediary layer between the vast information sources and the user's search queries. The extraction component acts as a mediator that processes diverse data sources, identifies relevant entities, and transforms them into structured results, making the detection and measurement of relevant resources easier despite the growing amount of information.
Solution Approach 2:
The system changes the parameter of information organization from unstructured web pages to structured entity mentions. By transforming information into standardized entity formats with associated metadata, the system makes relevant resources easier to detect and measure while maintaining coverage of diverse data sources.
3Extent of automation
If conventional search engines provide algorithmic results, then the search processing is automated, but the precision and accuracy of named entity retrieval deteriorate
Solution Approach 1:
The patent replaces conventional algorithmic search mechanisms with a named entity extraction and retrieval system. Instead of relying solely on keyword matching algorithms, the system uses extraction components that identify and retrieve specific named entities, providing more precise and accurate results while maintaining automated processing.
Data Source
AI summary
Architecture for finding related entities for web search queries. An extraction component takes a document as input and outputs all the mentions (or occurrences) of named entities such as names of people, organizations, locations, and products in the document, as well as entity metadata. An indexing component takes a document identifier (docID) and the set of mentions of named entities and, stores and indexes the information for retrieval. A document-based search component takes a keyword query and returns the docIDs of the top documents matching with the query. A retrieval component takes a docID as input, accesses the information stored by the indexing component and returns the set of mentions of named entities in the document. This information is then passed to an entity scoring and thresholding component that computes an aggregate score of each entity and selects the entities to return to the user.