Message Extractor Mapping for Template Adaptation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing email extraction systems require thousands of specialized extractors for different email templates, making it difficult to identify and use a suitable extractor when a merchant changes their email template, leading to inefficiencies and potential failure in extracting information from new templates.
Innovation Solution
A method that assigns a characteristic value to a received message using a mapping function, compares it to values associated with existing message extractors, and identifies a suitable extractor based on a similarity index and quality score, allowing for the creation or use of a new extractor to extract information from the message.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a specialized message extractor is trained for each specific email template, then extraction accuracy is improved, but the number of extractors required increases to thousands or hundreds of thousands
Solution Approach 1:
The patent creates a universal message extractor that can handle multiple email template types through characteristic value matching. Instead of having separate extractors for each template, a single extractor is designed to work with various templates by comparing characteristic values of incoming messages against stored template profiles, enabling one extractor to perform the function of thousands of specialized extractors
Solution Approach 2:
The patent transforms the approach by changing from template-specific extractors to a parameter-based matching system. Each email template is represented by characteristic values (parameters) that capture its essential features. The extractor uses these parameters to identify and process messages dynamically, allowing adaptation to different templates without creating new extractors
2Adaptability or versatility
If thousands of specialized extractors are maintained for different templates, then coverage of various email formats is improved, but difficulty in identifying a suitable extractor when templates change increases
Solution Approach 1:
The patent implements a feedback mechanism where the characteristic value of an incoming message is compared against stored template characteristic values. This comparison provides feedback that automatically identifies the matching template and selects the appropriate extractor configuration, eliminating the difficulty of manually or complexly identifying suitable extractors when templates change
Solution Approach 2:
The patent introduces characteristic values as an intermediary between email templates and extractors. Instead of directly matching templates to extractors, the system uses characteristic values as a mediator that simplifies the identification process. This intermediary layer enables automatic and efficient matching without the complexity of direct template-extractor correspondence
3Reliability
If a new extractor is created for each template change, then extraction functionality is maintained, but system reconfiguration effort and processing resources increase
Solution Approach 1:
The patent performs preliminary action by pre-calculating and storing characteristic values for known email templates. When a new message arrives, the system compares its characteristic value against the pre-stored values to quickly identify the matching template. This preliminary preparation eliminates the need for time-consuming reconfiguration when templates change, maintaining extraction functionality efficiently
Solution Approach 2:
The patent uses copying by creating characteristic value representations of email templates rather than copying the entire template structures or extractor configurations. This lightweight copying approach allows rapid identification and adaptation to new templates without the resource-intensive process of creating full extractor copies for each template variation
Data Source
AI summary
A computer system, and a method at a computer system, the method including applying a mapping function to a received message to create an characteristic value, wherein the mapping function is adapted to map similar messages to similar characteristic values; comparing the characteristic value to a value associated with each of a plurality of message extractors; determining that the characteristic value does not match any value associated with the plurality of message extractors; identifying at least one message extractor from the plurality of message extractors, the identifying determining that the value associated with the message extractor and the characteristic value from the received message, when compared, satisfy a similarity criterion; and using the identified at least one message extractor to extract information from the received message.