Behavior Graph Word Mining for Concealed Sensitive Words
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for mining sensitive words from the internet face challenges in accuracy and efficiency due to the high concealment of these words, which are often deliberately created to evade supervision.
Innovation Solution
A word mining method that constructs a behavior graph using search data, where first identification information, search sentences, and second identification information are nodes, and relationships between them are sides, allowing for the extraction of target words based on label vectors and preset labels, thereby improving mining accuracy and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional word mining methods are used to extract sensitive words from internet data, then the mining process can be performed with simple algorithms, but the mining accuracy is low due to the high concealment of sensitive words
Solution Approach 1:
The patent transforms the traditional single-dimension word mining approach into a multi-dimensional behavior analysis system. By constructing behavior graphs that incorporate user identification information, device identification information, search sentences, and temporal relationships, the system analyzes sensitive words from multiple dimensions simultaneously, significantly improving detection accuracy for concealed sensitive words
Solution Approach 2:
The patent segments the word mining process into distinct modular components: behavior data acquisition module, behavior graph construction module, label vector generation module, and target word extraction module. Each module handles a specific aspect of the analysis, making the complex system manageable and efficient
2Productivity
If traditional keyword matching methods are used for sensitive word detection, then the processing speed is fast, but the mining efficiency is low due to high concealment of sensitive words
Solution Approach 1:
The patent performs preliminary actions by pre-constructing behavior graphs from search data and pre-generating label vectors for different search sentences before actual sensitive word detection. This preprocessing enables faster real-time detection while maintaining high accuracy, as the computational heavy lifting is done in advance
Solution Approach 2:
The patent introduces behavior graphs and label vectors as intermediary structures between raw search data and final sensitive word detection results. These intermediaries capture complex behavioral patterns and relationships, enabling efficient and accurate detection without direct complex computation during the mining process
3Reliability
If simple keyword search is used to find sensitive words, then the method is easy to implement, but it cannot effectively identify concealed sensitive words created to evade supervision
Solution Approach 1:
The patent moves beyond simple text-based keyword matching by adding behavioral dimensions to the analysis. By incorporating user identification information, device identification information, search timing, and search patterns into behavior graphs, the system detects sensitive words based on behavioral anomalies rather than just keyword presence, significantly improving reliability against concealed sensitive words
Solution Approach 2:
The behavior graph construction module serves multiple functions simultaneously: it stores search data, models user behavior patterns, captures device characteristics, establishes temporal relationships, and provides the basis for label vector generation. This multi-functional approach improves detection reliability without proportionally increasing system complexity
Data Source
AI summary
A word mining method and apparatus, an electronic device and a readable storage medium are disclosed. The method includes: acquiring search data; taking first identification information, a search sentence and second identification information in the search data as nodes, and taking a relationship between the first identification information and the search sentence, a relationship between the first identification information and the second identification information and a relationship between the search sentence and the second identification information as sides to construct a behavior graph; obtaining a label vector of each search sentence in the behavior graph according to a search sentence with a preset label in the behavior graph; determining a target search sentence in the behavior graph according to the label vector; and extracting a target word from the target search sentence, and taking the target word as a word mining result of the search data.