POI Clustering Using Confidence Levels to Filter Noise
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for identifying points of interest (POI) from user track information are unreliable as they only consider historical location data and cannot accurately represent frequently visited locations that are important to the user, leading to low reliability and reference value.
Innovation Solution
A clustering method that acquires locating points within a preset period, generates stay points based on geographic proximity and time intervals, calculates confidence levels based on movement states, and clusters density-connected trusted stay points to form POIs, improving reliability and reference value by filtering out noise points.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If only historical location data is referred to for POI exploration, then the method is simple, but reliability and reference value of the explored POI are low
Solution Approach 1:
The patent segments the POI exploration process into multiple independent modules: stay point detection module, confidence level calculation module, and POI exploration module. Each module processes specific aspects of the data independently, improving reliability through systematic analysis while maintaining manageable complexity through modular design.
Solution Approach 2:
The patent introduces a new dimension of analysis by calculating confidence levels based on multiple factors including time intervals, spatial distances, and user behavior patterns. This multi-dimensional approach enhances POI reliability by considering additional parameters beyond simple location data.
2Reliability
If stay points are extracted by using spatial and temporal dimensions, then a single visit can be represented, but a POI place that is of great importance cannot be represented
Solution Approach 1:
The patent dynamically adjusts extraction parameters including time thresholds, spatial distance thresholds, and confidence level thresholds based on the specific characteristics of the data. This adaptive parameter adjustment enables accurate representation of important POI places while filtering out less significant locations.
Solution Approach 2:
The patent implements a feedback mechanism where confidence levels are calculated based on multiple factors including time intervals and spatial patterns, and this confidence information feeds back into the POI exploration process to improve the accuracy of identifying important locations.
3Reliability
If multiple locating points are clustered without confidence level filtering, then processing is faster, but noise points such as layover and jump points cannot be filtered out
Solution Approach 1:
The patent performs preliminary confidence level calculation and filtering before the final POI clustering process. By pre-identifying and filtering noise points such as layover and jump points based on confidence thresholds, the subsequent clustering operates on cleaner data, maintaining processing efficiency while improving data purity.
Solution Approach 2:
The patent extracts and removes noise points (layover points, jump points, pass-through points) from the locating point set before performing POI clustering. This extraction of harmful elements improves the purity of the final POI data while the selective filtering maintains reasonable processing efficiency.
Data Source
AI summary
A clustering method for a point of interest and a related apparatus are provided. The clustering method for a point of interest includes: acquiring a locating point set of a user within a preset period; generating a stay point set according to the locating point set, where each stay point in the stay point set represents one hot area; calculating a confidence level of each stay point in the stay point set; obtaining a trusted stay point from the stay point set by means of screening according to the confidence level of each stay point in the stay point set; and clustering density-connected trusted stay points to form a point of interest. By using technical solutions provided in the present disclosure, reliability and reference value of a POI can be effectively improved.