Binarized 3D Text Data Structure for Spam Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for detecting spam email, such as 'Bag-of-Words' and 'N-gram', struggle with character strings having high randomness and freedom, making it difficult to analyze and accurately classify spam, while 'distributed representation' faces challenges with using unpredetermined words.
Innovation Solution
A data structure is developed where text data is binarized on a per character type basis, converting it into three-dimensional data for feature learning and evaluation, allowing for the analysis of character strings with high randomness and enabling the application of deep learning and image processing methods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional methods like 'Bag-of-Words' or 'N-gram' are used for spam detection, then the analysis process is simple, but the accuracy is reduced due to inability to handle character strings with high randomness
Solution Approach 1:
The patent transforms text data from traditional two-dimensional representations (Bag-of-Words or N-gram matrices) into three-dimensional data structures by adding a character type dimension. This dimensional expansion allows the system to capture both the presence/absence of character types and their positional information, effectively handling random character strings while maintaining analytical capability
Solution Approach 2:
The patent changes the fundamental parameters of text representation by binarizing text data based on character types rather than whole words or N-grams. This parameter transformation enables the system to detect patterns in random character strings by focusing on character type distributions and positional relationships, thereby improving spam detection accuracy
2Adaptability or versatility
If text data is binarized on per character type basis to handle random strings, then the ability to analyze random character strings improves, but the data processing complexity increases
Solution Approach 1:
The patent segments text data into discrete character type units and represents them in a binarized three-dimensional structure. This segmentation approach breaks down complex random character strings into manageable character type components, making them analyzable while reducing the apparent complexity through systematic organization
Solution Approach 2:
The three-dimensional binarized data structure serves multiple functions: it captures character type presence/absence, preserves positional information, and handles random character strings. This multi-functional representation improves adaptability to various spam patterns without proportionally increasing processing complexity
3Measurement precision
If deep learning and image processing methods are applied to three-dimensional data, then the detection capability improves, but the computational resources required increase
Solution Approach 1:
The patent creates a three-dimensional binarized copy of the original text data that preserves essential spam detection features. This copied representation can be processed by deep learning and image processing algorithms, enabling sophisticated detection while allowing the original data to remain unchanged and efficiently stored
Data Source
AI summary
An evaluating method can be performed by a computer. The method includes acquiring two-dimensional data represented by a plurality of character types, converting the two-dimensional data to three-dimensional data by classifying the acquired two-dimensional data into a dimension of the plurality of character types, analyzing a feature of the three-dimensional data, and evaluating input data described in the plurality of character types based on the analyzed feature.