Browse Prior Art Database

POI Impact Evaluation Based on Crowd Flow Modeling

IP.com Disclosure Number: IPCOM000239363D
Publication Date: 2014-Nov-03
Document File: 5 page(s) / 96K

Publishing Venue

The IP.com Prior Art Database

Abstract

This disclosure proposes a system and a method that evaluate POI impact based on crowd flow modeling. In particular, a new interest value that is used to measure the interest degree of a POI of a person. This work solves the problem of the few training data is available for i) Behavior prediction against a certain person ; ii) Crowd behavior prediction against the place. The proposed system and method are scalable, and false tolerant for errors in the data sets

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 68% of the total text.

Page 01 of 5

POI Impact Evaluation Based on Crowd Flow Modeling

Problem background:


Given an area, evaluate the impact of adding a certain POI to that area. We need to determine the location of the new POI in the area to maximize the affection; The impact is modeled by the changing of the crowd flow. There are various applications such as Retailer Planning (store location, open & close & expanding) Government City Planning.

Technical problems:
i) Data Sparsity: Sparsity of the crowd data (i.e. POI paths), even worse, some area simply do not have crowd data
ii) How to model the affection of a POI? Need an objective metric instead of subjective judgment by the planner.

Technical solutions:
The system architecture is described as follows. In particular, it consists of two key components that this disclosure focuses on: crowd flow simulator, POI location planer.

For the crowd flow simulator, it runs simulation based on the following metric called crowd impact:
Given an area A which is a boundary of latitude and longitude, and the POI list P_A inside A, we try to introduce a new POI q into P, and find the maximum impact q could do to A.

where E(v(p)) is the expected daily crowd visits of p before q is added; E(v'(p)) is the expected daily crowd visits of p after q is added.

For the POI location planer, it generates testing POI locations which are fed into the crowd simulator for further simulation.

1


Page 02 of 5

The indexer and trainer is used to train a model as the input to the si...