Word Behavior Prediction Using Frequency Domain Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for predicting user behavior numbers for words in web sites face challenges in accuracy and reliability, particularly in distinguishing between stable and unstable sequences, leading to inefficient prediction algorithms and high operational complexity.
Innovation Solution
The method involves converting historical data sequences from the time domain to the frequency domain to identify estimated cycles and effective rate values, determining sequence stability, and using average value algorithms for stable sequences and main cycle-singularity-based predictions for unstable sequences, reducing operational complexity and improving prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional prediction algorithms are used for words with dramatic user behavior number changes, then prediction accuracy is improved, but the amount and complexity of operations as well as equipment consumption become quite great
Solution Approach 1:
The patent segments the prediction task by classifying words into different categories based on the stability of their user behavior number sequences. Stable sequences use simple average-based prediction, while unstable sequences use more complex algorithms. This segmentation allows each category to use appropriately matched prediction methods, avoiding unnecessary complexity for stable sequences while ensuring accuracy for unstable ones.
Solution Approach 2:
The patent changes the parameter of sequence stability (derived from standard deviation of user behavior numbers) to determine which prediction algorithm to apply. By monitoring this parameter, the system dynamically selects between simple and complex prediction methods, optimizing the balance between accuracy and computational resources.
2Measurement precision
If different prediction models are created for individual words, then prediction accuracy is improved, but it is impossible to create models as the number of words is extremely high
Solution Approach 1:
The patent creates universal prediction models for stable sequences that can be applied to any word exhibiting stable user behavior patterns. Instead of creating individual models for each word, the system identifies stable sequences and applies a general averaging model, making the prediction process scalable to extremely large numbers of words while maintaining efficiency.
Solution Approach 2:
The patent applies partial action by using simple average-based prediction for stable sequences rather than creating full complex models for all words. This partial application of sophisticated modeling only where necessary (for unstable sequences) dramatically reduces the overall computational burden while maintaining accuracy where it matters most.
3Measurement precision
If time sequence model is used to predict user behavior numbers, then prediction accuracy is improved for regular changes, but the sequence of user behavior number in practice generally does not meet the requirements
Solution Approach 1:
The patent changes the approach by first analyzing the stability parameter of the sequence before applying prediction methods. For stable sequences, it uses average-based prediction that doesn't require regular patterns. For unstable sequences, it applies time sequence models after identifying and handling singularities. This parameter-based adaptation makes the system compatible with various sequence types in practice.
Solution Approach 2:
The patent introduces sequence stability analysis as an intermediary step between raw data and prediction. This intermediary assessment determines whether the sequence is suitable for time sequence modeling or requires alternative approaches, bridging the gap between theoretical model requirements and practical data characteristics.
4Device complexity
If average value of user behavior numbers is used for prediction, then operation complexity is reduced, but prediction accuracy is poor for sequences with dramatic changes
Solution Approach 1:
The patent applies local quality by using different prediction methods tailored to local characteristics of each sequence. Stable sequences (with low variability) receive simple average-based prediction, while unstable sequences (with high variability or dramatic changes) receive more sophisticated time sequence model-based prediction. This localized approach ensures each sequence gets the appropriate level of complexity.
Solution Approach 2:
The patent makes the prediction system dynamic by automatically selecting between different prediction algorithms based on the stability characteristic of each sequence. The system adapts its complexity in real-time based on data characteristics, using simple methods when appropriate and complex methods when necessary, rather than applying a fixed approach to all sequences.
Data Source
AI summary
The present disclosure introduces a method, an apparatus and memory of predicting a user behavior number of a word for reducing the amount and the complexity of operation, saving the consumption of the equipment, and improving the accuracy and reliability of predictions. In an embodiment, a historical data sequence of the user behavior number of a word is converted from a time domain to a frequency domain. Based on the converted frequency domain, each estimated cycle and its effect rate value of the historical data sequence are ascertained. If the historical data sequence is stable, an average value of user behavior numbers of some historical data points before a prediction point is calculated as a user behavior number of the prediction point. Otherwise, the user behavior number is calculated based on a selected main cycle and a selected singularity.