Keyword Expansion for Financial Transaction Categorization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Taxpayers face laborious and time-consuming tasks when categorizing financial transactions due to complex tax regulations, leading to long user latencies despite advancements in software applications.
Innovation Solution
A method and system utilizing distributed-computing software to determine semantic relationships between terms, expand keyword collections, and calculate confidence intervals to assist users in categorizing transactions efficiently, reducing latency by suggesting expense categories based on user queries.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If users manually categorize financial transactions using complex tax regulation terminology, then categorization accuracy is maintained, but user time and effort increase significantly
Solution Approach 1:
The patent introduces an intermediary system consisting of keyword collections and semantic relationship analysis that mediates between the user's simple query and the complex categorization task. The system expands user queries using semantic relationships from a corpus and matches them against expense categories through confidence interval calculations, thereby maintaining accuracy while reducing user effort and time investment.
Solution Approach 2:
The patent performs preliminary actions by pre-computing keyword collections for each expense category and establishing semantic relationships between terms in advance. When a user submits a query, the system can quickly match it against pre-prepared keyword collections and confidence intervals, eliminating the need for users to manually navigate complex tax terminology and significantly reducing categorization time.
2Adaptability or versatility
If software applications provide comprehensive tax regulation terminology and category options, then categorization completeness is improved, but user interface complexity and search time increase
Solution Approach 1:
The patent extracts the essential keywords from comprehensive tax regulation terminology and organizes them into targeted keyword collections for each expense category. Instead of presenting users with the full complexity of tax regulation terminology, the system extracts and presents only the relevant keywords matched to user queries, thereby maintaining categorization completeness while significantly reducing interface complexity and search time.
3Measurement precision
If the system expands keyword collections using semantic relationships, then search accuracy is improved, but computational requirements and processing time increase
Solution Approach 1:
The patent performs the computationally intensive semantic relationship analysis and keyword collection expansion in advance, before users submit queries. The system pre-computes confidence intervals for keyword-category associations and stores these results. When users submit queries, the system only needs to perform simple matching operations against pre-computed data, thereby achieving high search accuracy while minimizing real-time computational requirements and processing time.
Data Source
AI summary
A method for category search. The method includes determining semantic relationships between terms in a corpus; obtaining, from an expense category hierarchy and for an expense category, a collection of keywords; and expanding the collection with a related keyword according to a semantic relationship of the relationships between the related keyword and a preexisting keyword in the collection. The method further includes extracting a segment from a description of a first historical transaction by a user of the financial product; and adding the extracted segment as an additional keyword to the collection when, for the extracted segment, the minimum of a first and a second confidence-interval bound calculated for a first transaction score and a first user score, respectively, satisfy a first threshold; and returning the name of the expense category in response to a user submitting a first query that comprises at least one the keyword in the collection.