Periodicity Detection in Noisy Click Streams
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional online support services lack robustness in detecting periodicity in click streams due to noisy network environments, as algorithms that rely on equally spaced clicks are not effective in handling temporal noise.
Innovation Solution
Assigning a periodicity score by forming an autocorrelation function from click streams and convolving it with a jitter kernel representative of the network's temporal noise distribution, allowing for the estimation of a period and confidence level in periodicity, thereby enhancing the detection of periodic patterns in noisy environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If conventional algorithms use direct measurement to find equally spaced clicks, then the detection method is simple, but the reliability of periodicity detection deteriorates in noisy network environments
Solution Approach 1:
The patent introduces an autocorrelation function as an intermediary between the raw click stream and the periodicity determination. This intermediary transforms the direct measurement approach into a statistical analysis approach, where the autocorrelation function processes the click timing data to reveal periodic patterns even in the presence of noise and jitter.
Solution Approach 2:
The patent changes the parameter being measured from direct time differences between clicks to the autocorrelation of time differences. By transforming the measurement parameter from raw timing to statistical correlation, the system can distinguish true periodicity from random noise in network environments.
2Ease of operation
If algorithms base determination solely on equally spaced clicks, then the measurement process is straightforward, but the measurement precision deteriorates due to network jitter
Solution Approach 1:
The autocorrelation function serves as a mediator that processes the raw click timing data, transforming it into a form where periodicity can be measured with higher precision. The intermediary computes the correlation of time differences with themselves, amplifying periodic signals while suppressing random jitter.
Solution Approach 2:
The autocorrelation function creates a copied and transformed version of the time difference data. By computing the correlation of the time differences with shifted versions of themselves, the system generates a new representation that reveals periodicity more precisely than direct measurement.
3Reliability
If the system accounts for temporal noise distribution through convolution, then the robustness of periodicity detection improves, but the computational complexity increases
Solution Approach 1:
The patent performs preliminary action by characterizing the network jitter through a probability distribution function before applying it to the periodicity detection. This pre-characterization of the noise model allows the convolution to efficiently account for temporal variations without requiring complex real-time adjustments.
Solution Approach 2:
The probability distribution function of network jitter acts as an intermediary that models the temporal noise. By convolving this jitter model with the autocorrelation function, the system mediates between the raw data and the periodicity determination, accounting for network variations in a systematic way.
Data Source
AI summary
Sequences of discrete events, such as clicks on a website, are evaluated for periodic behavior, a period is calculated, and the sequence is scored to determine the confidence that the sequence really exhibits periodicity. The random variations on the timing of the discrete events due to transmission delays or other factors may be reduced or eliminated from the evaluation. An apparatus for performing the method of evaluation may include a computer programmed to carry out the method.