Resident area prediction method, apparatus, device, and storage medium

Patent metadata

Field	Value
Publication number	US-11829447-B2
Application number	US-202117173142-A
Country	US
Kind code	B2
Filing date	Feb 10, 2021
Priority date	Sep 29, 2020
Publication date	Nov 28, 2023
Grant date	Nov 28, 2023

Abstract

This disclosure discloses a resident area prediction method, apparatus, device and storage medium, involving artificial intelligence technology, big data, deep learning and multi-task learning. The specific implementation plan is: acquiring a resident area data of a target user, and the resident area data including the resident area of the target user and the corresponding resident time; obtaining an association relationship between the resident areas of the target user by inputting the resident area data into an area relationship model, and the area relationship model is used to reflect a position relationship between the areas; determining a time-sequence relationship between the areas visited by the target user, according to the association relationship, the resident time and the visiting POI data; predicting a target resident area of the target user, according to the time-sequence relationship and the basic attribute information of the target user.

First claim

What is claimed is:

1. A resident area prediction method, comprising:
acquiring resident area data and visiting point of interest (POI) data of a target user, wherein the resident area data comprises resident areas and corresponding resident times of the target user;
inputting the resident area data into an area relationship model to obtain an association relationship between the resident areas of the target user, wherein the area relationship model is configured to reflect a position relationship between areas;
determining a time-sequence relationship between areas visited by the target user, according to the association relationship, the resident times and the visiting POI data; and
predicting a target resident area of the target user, according to the time-sequence relationship and basic attribute information of the target user.

2. The method according to claim 1 , wherein the inputting the resident area data into an area relationship model to obtain an association relationship between the resident areas of the target user comprises:
obtaining resident area association relationships at different times, according to the resident areas and the corresponding resident times;
obtaining association relationships of different resident areas at different times, according to the area relationship model; and
obtaining the association relationship between the resident areas of the target user, according to the resident areas at different times.

3. The method according to claim 1 , wherein the determining a time-sequence relationship between areas visited by the target user, according to the association relationship, the resident times and the visiting POI data, comprises:
performing a multi-source information fusion processing on the association relationship, the resident times, and the visiting POI data to obtain fused information; and
inputting the fused information into a time-sequence relationship model to determine the time-sequence relationship between the areas visited by the target user, wherein the time-sequence relationship model is configured to reflect a time-sequence relationship of user migration behaviors.

4. The method according to claim 3 , wherein the performing a multi-source information fusion processing on the association relationship, the resident times, and the visiting POI data to obtain fused information comprises:
performing a pooling processing on the resident times and the visiting POI data respectively to obtain an intermediate representation vector; and
performing a vector splicing processing on the intermediate representation vector and the association relationship to obtain the fused information.

5. The method according to claim 1 , wherein the predicting a target resident area of the target user, according to the time-sequence relationship and basic attribute information of the target user, comprises:
performing an aggregation processing on the time-sequence relationship and the basic attribute information of the target user;
obtaining resident probabilities of the target user in different areas according to a result of the aggregation processing and a task for resident area prediction; and
predicting the target resident area of the target user according to the resident probabilities of the target user in different areas.

6. The method according to claim 5 , wherein the performing an aggregation processing on the time-sequence relationship and the basic attribute information of the target user comprises:
converting the basic attribute information of the target user into a numeric value to obtain basic attribute data corresponding to the basic attribute information; and
performing an aggregation processing on the time-sequence relationship and the basic attribute data.

7. The method according to claim 1 , further comprising:
performing an aggregation processing on the time-sequence relationship and the basic attribute information of the target user;
obtaining a visiting intention probability of the target user in the target resident area, according to a result of the aggregation processing and a task for user visiting intention prediction; and
determining the visiting intention of the target user in the target resident area, according to visiting intention probabilities of the target user corresponding to different categories of points of interest (POIs) in the target resident area.

8. The method according to claim 1 , wherein the acquiring resident area data and visiting POI data of a target user comprises:
converting resident area information and visiting POI information of the target user into numeric values respectively to obtain the resident area data corresponding to the resident area information and the visiting POI data corresponding to the visiting POI information.

9. The method according to claim 1 , wherein the area relationship model is constructed according to resident areas of a user.

10. The method according to claim 9 , wherein
the area relationship model is further constructed by modeling a dynamic association relationship between the resident areas based on a dual-path graph convolutional network DGCN.

11. An apparatus for constructing an area relationship model, comprising: at least one processor and a memory communicatively connected with the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the method according to claim 9 .

12. A non-transitory computer-readable storage medium storing computer instructions for causing a computer to execute the method according to claim 1 .

13. A resident area prediction apparatus, comprising: at least one processor and a memory communicatively connected with the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to:
acquire resident area data and visiting point of interest (POI) data of a target user, wherein the resident area data comprises resident areas and corresponding resident times of the target user;
input the resident area data into an area relationship model to obtain an association relationship between the resident areas of the target user, wherein the area relationship model is configured to reflect a position relationship between areas;
determine a time-sequence relationship between areas visited by the target user, according to the association relationship, the resident times and the visiting POI data; and
predict a target resident area of the target user, according to the time-sequence relationship and basic attribute information of the target user.

14. The apparatus according to claim 13 , wherein the instructions are executed by the at least one processor to enable the at least one processor to:
obtain resident area association relationships at different times, according to the resident areas and the corresponding resident times;
obtain association relationships of different resident areas at different times, according to the area relationship model; and
obtain the association relationship between the resident areas of the target user, according to the resident areas at different times.

15. The apparatus according to claim 13 , wherein the instructions are executed by the at least one processor to enable the at least one processor to:
perform a multi-source information fusion processing on the association relationship, the resident times, and the visiting POI data to obt

Assignees

• Beijing Baidu Netcom Sci & Tech Co Ltd

Inventors

• Lu Xinjiang
• Zhu Nengjun
• Xiong Hui

Classifications

• G06N3/09

  Supervised learning · CPC title

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06N3/0442

  characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU] · CPC title

• G06F18/2323 Primary

  based on graph theory, e.g. minimum spanning trees [MST] or graph cuts · CPC title

• G06F17/18

  for evaluating statistical data {, e.g. average values, frequency distributions, probability functions, regression analysis (forecasting specially adapted for a specific administrative, business or logistic context G06Q10/04)} · CPC title