Search Ranking via NLP Feature Scoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional search systems rely heavily on keywords, leading to reduced quality and accuracy in search results due to equal weighting of all keywords, regardless of context, resulting in irrelevant matches and decreased relevancy and ranking.
Innovation Solution
A method and system that utilize natural language processing (NLP) to derive document features based on query features, calculate feature scores, and prioritize document features using pre-defined rules to rank search results, enhancing relevancy and accuracy by emphasizing important tokens and metadata.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional search systems use keyword matching with equal weighting for all keywords, then the search process is simple and fast, but the relevancy and accuracy of search results deteriorates
Solution Approach 1:
The patent applies local quality by assigning different weights to different keywords based on their importance to the query intent. Instead of treating all keywords equally, the system identifies and prioritizes important tokens (keywords) that carry more semantic weight, while reducing the influence of less important tokens. This differential weighting approach improves search result relevancy without requiring complete system redesign.
Solution Approach 2:
The patent changes the parameter of keyword weighting from uniform (equal weight for all keywords) to variable (different weights based on importance). By dynamically adjusting the weight parameter for each keyword based on its significance to the query, the system achieves better search accuracy while maintaining computational feasibility through efficient scoring mechanisms.
2Quantity of substance
If search systems prioritize matching non-important tokens, then more documents are retrieved, but the quality and accuracy of search results deteriorates
Solution Approach 1:
The patent applies local quality by differentiating between important and non-important tokens in the query. Instead of treating all tokens uniformly, the system identifies important tokens that are critical to query intent and assigns them higher weights. This ensures that documents matching important tokens are prioritized, improving accuracy while still considering non-important tokens for broader document retrieval.
Solution Approach 2:
The patent introduces asymmetry in the treatment of different tokens by assigning unequal weights based on their importance. Important tokens receive higher weightage in the scoring mechanism, while non-important tokens receive lower weightage. This asymmetric approach allows the system to retrieve a sufficient quantity of documents while maintaining high accuracy by emphasizing matches on critical tokens.
3Productivity
If search systems provide equal weightage to all keywords, then the processing is simpler and faster, but the relevancy of search results deteriorates
Solution Approach 1:
The patent applies local quality by assigning different weights to different keywords based on their importance to the query intent. Instead of treating all keywords equally, the system identifies and prioritizes important tokens (keywords) that carry more semantic weight, while reducing the influence of less important tokens. This differential weighting approach improves search result relevancy without requiring complete system redesign.
Solution Approach 2:
The patent changes the parameter of keyword weighting from uniform (equal weight for all keywords) to variable (different weights based on importance). By dynamically adjusting the weight parameter for each keyword based on its significance to the query, the system achieves better search accuracy while maintaining computational feasibility through efficient scoring mechanisms.
Data Source
AI summary
This disclosure relates to method and system for improving relevancy and ranking of a search result for a search query. The method may include accessing documents of the search result. Each document may be associated with derived document features that may be derived based on an evaluation of document features with respect to query features. For each document, the method may further include determining a feature score for each of the derived document features based on one or more of the query features and initializing each element of an order array with the feature score for each ordered document feature. The ordered document features may be the derived document features that are ordered based on pre-defined rules. The method may further include ranking the documents by comparing and ranking the order array for each of the documents and presenting an updated search result based on the ranking of the documents.