Message Sorting System Using Segmented Extraction and Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing message sorting systems using machine-learning devices often fail to achieve sufficient accuracy in categorizing messages, particularly in representing the opinions of posting persons, leading to inaccurate analysis results.
Innovation Solution
A message sorting system that includes an extraction block to filter messages based on posting person, reply destination, or content rules, followed by a machine-learning sorting block to accurately categorize the extracted messages.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If all messages are sorted by a machine-learning sorting device, then the sorting process is automated, but the sorting accuracy is insufficient and messages not representing opinions are incorrectly categorized
Solution Approach 1:
The message sorting process is segmented into two distinct stages: first, an extraction unit filters messages based on predetermined rules (posting person, reply destination, contents) to isolate candidate messages representing opinions; second, a machine-learning sorting device categorizes only these extracted messages. This segmentation allows the machine-learning device to focus on a refined subset, improving accuracy while maintaining automation.
Solution Approach 2:
Before the machine-learning sorting device processes messages, a preliminary extraction action is performed using predetermined rules to filter and select messages that are likely to represent posting persons' opinions. This preliminary filtering action prepares the data in advance, enabling the machine-learning device to operate more effectively on pre-qualified messages, thereby improving sorting accuracy.
2Measurement precision
If message filtering rules are applied before machine-learning sorting, then sorting accuracy improves, but the system complexity increases
Solution Approach 1:
The system is segmented into two functional units: an extraction unit implementing predetermined rules and a machine-learning sorting device. This segmentation distributes complexity across modular components, where each unit has a specific, well-defined function. The extraction unit handles rule-based filtering while the machine-learning unit handles categorization, making the overall system complexity manageable through functional decomposition.
Solution Approach 2:
The extraction unit acts as an intermediary between the raw message input and the machine-learning sorting device. It applies predetermined rules to filter and prepare messages before they reach the machine-learning device, serving as a mediator that simplifies the input data and improves the effectiveness of the subsequent sorting operation.
Data Source
AI summary
A message sorting systems includes: an extraction block configured to extract some of a plurality of posted messages on the basis of a rule with respect to message posting person, reply destination, or contents; and a sorting block configured to sort the extracted messages through a machine-learning sorting device. A message sorting method includes: extracting some of a plurality of posted messages on the basis of a rule with respect to message posting person, reply destination, or contents; and sorting the extracted messages through a machine-learning sorting device.