Search Ranking via Edit Distance and TAUC Neural Network
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current search engine ranking methods are inefficient in providing accurate relevance of search results, as they rely on traditional ranking functions that do not effectively handle near-matches or compound terms, leading to suboptimal user satisfaction.
Innovation Solution
The proposed solution involves computing an edit distance between a query string and document information (TAUC: title, anchor text, URL, and clicks) using a neural network, which splits compound terms and filters anchor text at index time to improve relevance ranking by detecting near-matches and assigning relevance scores based on edit distance operations (insertion, deletion, and position costs).
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional ranking functions are used to rank candidate documents, then the search engine can process queries quickly, but the relevance accuracy of search results deteriorates
Solution Approach 1:
The patent segments the ranking process into multiple stages: first using traditional ranking functions for quick initial sorting, then applying edit distance calculations and neural network-based relevance metrics to top-ranked candidates. This segmentation allows the system to maintain processing speed while improving relevance accuracy through progressive refinement.
Solution Approach 2:
The patent performs preliminary actions by pre-computing document features, anchor text profiles, and proximity relationships during indexing. This preliminary preparation enables faster query processing while maintaining high relevance accuracy, as the expensive computational work is done in advance rather than at query time.
2Reliability
If the candidate document set includes all documents containing query keywords, then completeness is improved, but the ranking efficiency deteriorates due to large cardinality
Solution Approach 1:
The patent applies local quality by differentiating the processing approach for different parts of the candidate set. High-quality documents (those with strong keyword matches and relevant metadata) receive more sophisticated analysis including edit distance and neural network evaluation, while lower-priority candidates receive simpler filtering. This selective approach maintains completeness for important results while improving overall efficiency.
3Adaptability or versatility
If compound terms in URLs are not split, then the indexing process is simpler, but the discovery of query terms deteriorates
Solution Approach 1:
The patent segments compound terms in URLs into individual queryable units during indexing. For example, 'companystore' is split into 'company' and 'store', allowing the search engine to discover and match individual query terms within compound terms. This segmentation improves query term discovery while the segmented index structure maintains reasonable processing efficiency.
4Measurement precision
If anchor text is not filtered, then all potential matches are considered, but the ranking precision deteriorates due to noise
Solution Approach 1:
The patent applies partial action by filtering anchor text to retain only the top N most relevant anchors for each document, rather than processing all anchors. This selective filtering removes noise from less relevant anchors while preserving the most informative ones, improving ranking precision without losing significant match information. The parameter N is tuned to balance between precision and information retention.
Data Source
AI summary
Architecture for extracting document information from documents received as search results based on a query string, and computing an edit distance between the data string and the query string. The edit distance is employed in determining relevance of the document as part of result ranking by detecting near-matches of a whole query or part of the query. The edit distance evaluates how close the query string is to a given data stream that includes document information such as TAUC (title, anchor text, URL, clicks) information, etc. The architecture includes the index-time splitting of compound terms in the URL to allow the more effective discovery of query terms. Additionally, index-time filtering of anchor text is utilized to find the top N anchors of one or more of the document results. The TAUC information can be input to a neural network (e.g., 2-layer) to improve relevance metrics for ranking the search results.