Vector Matching Engine for Heterogeneous Text Collections
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data processing techniques for matching a set to evaluate with a reference list are inadequate as they only consider character strings independently and require weighting based on a collection of documents, making them unsuitable for evolving and heterogeneous collections, and fail to account for groups of character strings effectively.
Innovation Solution
A method that calculates a distance between a reference vector and a vector representing the set to evaluate, considering character strings and groups, and uses centrality coefficients to weight matching scores, providing an ordered list of reference list entries based on their relevance, which includes calculating triplet lists and combining distances and centrality coefficients to determine matching scores.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If character strings are weighted based on a collection of documents, then matching can be performed, but it becomes ill-suited for managing evolving and heterogeneous collections
Solution Approach 1:
The patent extracts the weighting mechanism from the collection context. Instead of weighting character strings based on their distribution across a collection of documents, the method calculates weights based solely on the internal structure and relationships within the single document being processed. This extraction allows the matching system to adapt to evolving and heterogeneous collections without requiring reweighting based on collection statistics.
Solution Approach 2:
The patent introduces dynamic weighting where the weights of character strings are calculated based on their specific relationships and groupings within each document rather than static collection-wide frequencies. This dynamic approach allows the system to adapt to changing document collections while maintaining consistent matching behavior, as weights are recomputed based on actual document structure rather than fixed collection statistics.
2Ease of manufacture
If character strings are considered independently, then processing is simplified, but groups of character strings are not accounted for effectively
Solution Approach 1:
The patent merges the processing of character strings by introducing group-based weighting. Instead of treating each character string independently, the method identifies groups of character strings that appear together in documents and assigns weights to these groups. This merging allows the system to capture contextual relationships and phrase-level patterns while maintaining a relatively simple processing framework based on vector space models.
3Extent of automation
If supervised learning is used for multi-label classification, then documents can be indexed, but a large quantity of manually indexed documents is required
Solution Approach 1:
The patent implements self-service by enabling the system to automatically learn from the structure and content of documents without requiring manual indexing. The method calculates weights and similarities based on the inherent patterns within documents themselves, allowing the system to autonomously perform multi-label classification and indexing. This eliminates the need for large quantities of manually indexed documents while maintaining high automation levels.
Data Source
AI summary
A method of matching a set to be evaluated and a reference list, the reference list being associated with a reference vector representative of the entries in the list. Such a method of matching includes: calculating a distance between the reference vector and a vector, associated with the set to evaluate, representative of elements contained in the set to evaluate, the elements comprising character strings and groups of character strings; for each entry in the reference list, calculating a first matching score for the set to evaluate and for the entry in the reference list, on the basis of the distance calculated between the reference vector and the vector associated with the set to evaluate; providing a list of entries from the reference list ordered according to the first calculated matching scores.