Criteria-Specific Authority Ranking for Search Relevance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional information retrieval systems become inefficient with large, loosely structured, or unstructured data repositories, such as the World Wide Web, as they deliver numerous search results with similar authority, failing to adequately reflect user context and diverse information needs.
Innovation Solution
The system analyzes and scores individual nodes in an interlinked data repository based on context-specific criteria, weighting links by intrinsic node scores to propagate authority in a criteria-specific manner, thereby enhancing the relevance of search results to user interests and needs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional information retrieval systems are used to search large data repositories, then the system can return all documents containing the search token, but the system becomes inefficient and delivers numerous search results with similar authority, making it difficult for users to identify relevant information
Solution Approach 1:
The patent changes the parameter of link weighting from uniform to criteria-specific weights. Each link is assigned a weight based on criteria such as link type, source node properties, and target node properties, transforming the ranking mechanism from simple authority counting to weighted authority propagation, thereby improving search result relevance
Solution Approach 2:
The patent applies local quality by assigning different weighting criteria to different links based on their specific characteristics. Instead of treating all links equally, the system evaluates each link individually using appropriate criteria (e.g., editorial links receive different weights than user-generated links), enabling more precise authority measurement for each node
2Adaptability or versatility
If generic authority ranking methods are used that give each link equal weight, then the ranking is simple to compute, but the ranking fails to reflect user context and diverse information needs
Solution Approach 1:
The patent introduces dynamics by making the link weighting adaptive rather than static. The weighting criteria can be adjusted based on user context, search query characteristics, and node properties, allowing the ranking system to dynamically adapt to different information needs and contexts while maintaining a structured computational framework
Solution Approach 2:
The patent segments the authority propagation process into multiple independent weighting criteria that can be applied separately to different link types and node properties. This segmentation allows the complex ranking algorithm to be broken down into manageable components, each handling specific aspects of authority evaluation
3Quantity of substance
If the number of search results with similarly high authority increases due to Web growth, then more comprehensive coverage is achieved, but the number of results exceeds what a human user could reasonably review
Solution Approach 1:
The patent changes the differentiation parameter by introducing weighted criteria that distinguish between different types of links and nodes. Even when multiple nodes have high authority, the weighted criteria (link type, source/target properties) create finer-grained distinctions, enabling better result differentiation without reducing coverage
Data Source
AI summary
In one embodiment, a method of ranking a plurality of linkable documents accessible via a computer network is disclosed. A server analyzes each document so as to measure one or more criteria of the document, each criterion being independent of links between the document and other documents accessible via the computer network and at least one criterion relating to the presence of unsafe or inappropriate content in the corresponding document. Further, the server assigns an independent intrinsic score to each document based on the analysis. The server identifies for each document all links between the respective document and one or more other documents. Further, the server generates a ranking score for each document based on the intrinsic scores and iteratively updating the ranking score of each document based on previous ranking scores and the intrinsic scores of documents linked therewith.