Method for determining competitive relation of points of interest, device

Patent metadata

Field	Value
Publication number	US-11789985-B2
Application number	US-202117480780-A
Country	US
Kind code	B2
Filing date	Sep 21, 2021
Priority date	Sep 25, 2020
Publication date	Oct 17, 2023
Grant date	Oct 17, 2023

Abstract

A method for determining a competitive relation of Points of Interest (POI), and a device are provided in the present disclosure. The specific implementation includes: determining POI representation data between two target POIs based on service-related data of the target POIs; and determining a competitive relation between the target POIs based on the POI representation data.

First claim

What is claimed is:

1. A method for determining a competitive relation of Points of Interest (POI), comprising:
determining service-related data of target POIs based on a pre-constructed POI heterogeneous information network, wherein the POI heterogeneous information network is constructed by: extracting service keywords of comment data of a plurality of candidate POIs; and constructing the POI heterogeneous information network based on co-occurrence relations of respective candidate POIs and similarity relations of the service keywords;
determining POI representation data between the target POIs based on the service-related data of the target POIs; and
determining a competitive relation between the target POIs based on the POI representation data,
wherein constructing the POI heterogeneous information network based on the co-occurrence relations of respective candidate POIs and the similarity relations of the service keywords comprises:
constructing a service relation edge between keyword nodes of the service keywords based on the similarity relation of the service keywords;
constructing a first relation edge between a POI node of the candidate POI and the keyword node of the service keyword based on a consistency of paradigmatic points to which the candidate POI and the service keyword belong; and
determining a service edge attribute of each service relation edge based on a similarity between the service keywords associated with the service relation edge, wherein the similarity between the service keywords indicates a semantic similarity between the service keywords;
wherein determining the service-related data of the target POIs based on the pre-constructed POI heterogeneous information network comprises:
determining target keyword nodes associated with the target POIs based on the first relation edge and the service relation edge in the POI heterogeneous information network; and
determining service-related data of a service type of the target POIs based on the service edge attribute of the service relation edge between the target keyword nodes,
wherein determining the POI representation data between the target POIs based on the service-related data of the target POIs comprises:
determining a POI service representation of each target POI based on the service-related data of the service type, by adopting a feature extraction network to extract the POI service representation in the service-related data of each target POI; and
determining the POI representation data between the target POIs based on the POI service representation of each target POI, by introducing an average fully connected layer to process the POI service representation of each target POI.

2. The method of claim 1 , wherein constructing the first relation edge between the POI node of the candidate POI and the keyword node of the service keyword based on the consistency of the paradigmatic points to which the candidate POI and the service keyword belong comprises:
constructing a first sub-relation edge between the POI node and a paradigmatic point node of the paradigmatic point based on an ownership of the candidate POI and the paradigmatic point; and
constructing a second sub-relation edge between a paradigmatic point node of the paradigmatic point and the keyword node based on an ownership of the paradigmatic point and service keyword.

3. The method of claim 2 , further comprising:
constructing a paradigmatic relation edge between paradigmatic point nodes of paradigmatic points to which the candidate POIs belong based on the co-occurrence relation between the candidate POIs;
determining a paradigmatic edge attribute of each paradigmatic relation edge based on connected data between the paradigmatic point nodes connected by the paradigmatic relation edge; and
determining a second edge attribute of each second sub-relation edge based on a contribution degree of the service keyword associated with the second sub-relation edge to paradigmatic comment data of the paradigmatic point associated with the second sub-relation edge;
wherein determining the service-related data of the target POIs based on the pre-constructed POI heterogeneous information network further comprises:
respectively determining target paradigmatic point nodes and target keyword nodes associated with the target POIs based on the first sub-relation edge and the second sub-relation edge;
determining service-related data of a paradigmatic type of the target POIs based on the paradigmatic edge attribute between the target paradigmatic point nodes; and
determining service-related data of a dissimilar node type of the target POIs based on the second edge attribute between the paradigmatic point node and the target keyword node.

4. The method of claim 3 , wherein determining the POI representation data between the target POIs based on the service-related data of each target POI comprises:
determining a POI service representation of each target POI based on the service-related data of the service type and the service-related data of the dissimilar node type;
determining a POI paradigmatic representation of each target POI based on the service-related data of the paradigmatic type and the service-related data of the dissimilar node type;
using the POI paradigmatic representation of one target POI to update the POI service representation of the other target POI; and
determining the POI representation data between the target POIs based on the POI service representation of each target POI.

5. An electrical device, comprising:
at least one processor, and
a memory communicatively coupled to the at least one processor; wherein,
the memory is configured to store instructions executable by at least one processor; when the instructions are executed by the at least one processor, the at least one processor is caused to execute the method for determining a competitive relation of POIs, comprising:
determining service-related data of target POIs based on a pre-constructed POI heterogeneous information network, wherein the POI heterogeneous information network is constructed by: extracting service keywords of comment data of a plurality of candidate POIs; and constructing the POI heterogeneous information network based on co-occurrence relations of respective candidate POIs and similarity relations of the service keywords;
determining POI representation data between the target POIs based on the service-related data of the target POIs; and
determining a competitive relation between the target POIs based on the POI representation data,
wherein constructing the POI heterogeneous information network based on the co-occurrence relations of respective candidate POIs and the similarity relations of the service keywords comprises:
constructing a service relation edge between keyword nodes of the service keywords based on the similarity relation of the service keywords;
constructing a first relation edge between a POI node of the candidate POI and the keyword node of the service keyword based on a consistency of paradigmatic points to which the candidate POI and the service keyword belong; and
determining a service edge attribute of each service relation edge based on a similarity between the service keywords associated with the service relation edge, wherein the similarity between the service keywords indicates a semantic similarity between the service keywords;
wherein determining the service-related data of the target POIs based on the pre-constructed POI heterogeneous information network comprises:
determining target keyword nodes associated with the target POIs based on the first relation edge and the service relation edge in the POI heterogeneous information network; and
determining service-related data of a service type of the target POIs based on the service edge

Assignees

• Beijing Baidu Netcom Sci & Tech Co Ltd

Inventors

• Li Shuangli
• Zhou Jingbo
• Xu Tong
• Xiong Hui

Classifications

• G06F16/29 Primary

  Geographical information databases · CPC title

• G06F16/9536

  Search customisation based on social or collaborative filtering · CPC title

• G06F16/9537 Primary

  Spatial or temporal dependent retrieval, e.g. spatiotemporal queries · CPC title

• G06F16/9535

  Search customisation based on user profiles and personalisation · CPC title

• G06F40/216

  using statistical methods · CPC title