Cross-Domain Document Ranking via Multi-Property Descriptors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current document ranking techniques in computing systems, such as those used in asset reuse applications, face challenges in accurately selecting relevant solution documents from large corpora, leading to low precision and recall, which increases resource consumption and requires additional search requests.
Innovation Solution
A computer-implemented method generates descriptors for documents with corresponding values and confidence indexes for multiple properties, allowing for the comparison and ranking of documents across different domains, thereby identifying relevant solution documents more effectively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional bag-of-words ranking techniques are used, then the ranking process is simple and fast, but the precision and recall of selecting relevant solution documents are low
Solution Approach 1:
The patent segments documents into multiple properties (e.g., title, abstract, body, metadata) and generates separate descriptors for each property. This segmentation allows the system to evaluate different aspects of document relevance independently, thereby improving precision without requiring a completely complex overhaul of the ranking system.
Solution Approach 2:
The patent introduces a new dimension to document comparison by generating descriptors that capture semantic meaning beyond simple word matching. This includes analyzing document structure, property relationships, and contextual information, adding dimensional depth to the ranking process that improves precision while managing complexity through structured approaches.
2Measurement precision
If additional search requests are submitted to find relevant solution documents, then the recall is improved, but the computational resources and network traffic increase
Solution Approach 1:
The patent performs preliminary actions by generating comprehensive descriptors for all documents in the corpus beforehand, including extracting and analyzing multiple properties and their relationships. This preliminary processing enables more accurate single-pass ranking, improving recall without requiring multiple search requests and reducing computational resource consumption during actual search operations.
Solution Approach 2:
The patent replaces the mechanical approach of submitting multiple search requests with a more sophisticated descriptor-based comparison system. By using semantic descriptors and confidence indexes, the system can achieve better recall in a single search operation, substituting iterative mechanical searching with a more efficient analytical approach.
3Measurement precision
If multiple properties and confidence indexes are analyzed for each document, then the ranking quality is improved, but the processing time and computational complexity increase
Solution Approach 1:
The patent applies local quality by assigning different levels of analysis and confidence indexes to different document properties based on their relevance. Not all properties are processed with equal depth - the system selectively analyzes properties that are more critical for the specific search context, improving ranking quality while reducing overall processing time by avoiding uniform deep analysis of all properties.
Data Source
AI summary
A solution is proposed for ranking documents belonging to two different domains. A corresponding method comprises generating a descriptor for each of the documents; the descriptor comprises corresponding values and confidence indexes of multiple properties (of the corresponding document); the documents of a domain are ranked with respect to a document of another domain according to a comparison of their descriptors. A computer program product for performing the method are also proposed. Moreover, a computing system for implementing the method is proposed.