System and method for processing point-of-interest data

Patent metadata

Field	Value
Publication number	US-2021390392-A1
Application number	US-202017082874-A
Country	US
Kind code	A1
Filing date	Oct 28, 2020
Priority date	Jun 15, 2020
Publication date	Dec 16, 2021
Grant date	—

Abstract

Methods employing temporal and/or cultural data are provided for automatically assigning one or more semantic tags to a point-of-interest (POI) using a processor. The POI is represented by attribute data. A dataset including temporal attribute data and/or cultural data is provided to a multilabel classifier comprising a neural network model. One or more predicted semantic tags for the POI are received from an output of the multilabel classifier. The predicted semantic tags are stored in a database as additional attribute data of the POI.

First claim

1 . A method for automatically assigning one or more semantic tags to a point-of-interest (POI) using a processor, the POI being represented by attribute data, the method comprising:
receiving the attribute data for the POI, the attribute data comprising temporal attribute data and one or more of spatial attribute data or metadata; providing the received attribute data to a multilabel classifier comprising a neural network model; receiving one or more predicted semantic tags for the POI from an output of the multilabel classifier; and storing the predicted semantic tags in a database as additional attribute data of the POI.

2 . The method of claim 1 , wherein the predicted semantic tags comprise category labels for the POI;
wherein the category labels are taken from a label set stored in the database.

3 . The method of claim 2 , wherein the metadata comprises one or more observed semantic tags for the POI;
wherein the one or more semantic tags comprise category labels taken from the label set.

4 . The method of claim 1 , wherein the metadata comprises a unique identifier for the POI.

5 . The method of claim 1 , wherein the spatial attribute data comprises geospatial data.

6 . The method of claim 1 , wherein the temporal data comprises one or more of opening times, closing times, or access times for the POI.

7 . The method of claim 1 , wherein said providing the received attribute data to a multilabel classifier comprises:
vectorizing the received attribute data; concatenating the vectorized data; and inputting the concatenated data to the multilabel classifier.

8 . The method of claim 7 , wherein the received attribute data comprises at least one categorical variable; and
wherein said vectorizing comprises representing the at least one categorical variable by one-hot encoding.

9 . The method of claim 7 , wherein the received attribute data comprises at least one sequential variable; and
wherein said vectorizing comprises processing the at least one sequential variable using an n-gram character-based long short-term memory (LSTM) model.

10 . The method of claim 7 , wherein the received attribute data comprises at least one spatial variable;
wherein said vectorizing comprises one or more of:
modeling the at least one spatial variable using a discretized input space; or
mapping the at least one spatial variable to a geographical region represented by a categorical variable or a sequential variable.

11 . The method of claim 7 , wherein said vectorizing comprises one or more of:
representing the temporal data as a categorical variable using one-hot encoding; representing the temporal data as a periodic variable by transforming the temporal data into one or more vectors respectively representing dimensions of the temporal data; or representing the temporal data as a formatted string using an n-gram character based long short-term memory (LSTM) model.

12 . The method of claim 1 , wherein the predicted semantic tags comprise category labels for the POI;
wherein the category labels are taken from a label set stored in the database; and wherein the received one or more predicted semantic tags comprises a probability score for each of the category labels in the label set.

13 . The method of claim 12 , further comprising:
selecting the received one or more predicted semantic tags for which the probability score meets or exceeds a threshold.

14 . The method of claim 1 , wherein the multilabel classifier comprises a neural network model.

15 . The method of claim 1 , wherein the method further comprises:
providing a training set, wherein said providing comprises:
filtering attribute data for each of a plurality of POIs, wherein the filtered attribute data comprises the temporal attribute data and metadata, the metadata comprising at least one semantic tag;
for each of the plurality of POIs, selecting the at least one semantic tag to be completed as a target; and
vectorizing a remainder of the filtered attribute data to provide training data for the training set; and
training the multilabel classifier using the provided training set.

16 . The method of claim 15 , wherein the filtered attribute data for each of the plurality of POIs comprises POI names;
wherein the POI names among the plurality of POIs are represented by a plurality of languages and/or scripts.

17 . The method of claim 1 , wherein the database comprises a global crowdsourced database.

18 . The method of claim 1 , wherein the attribute data provided to the multilabel classifier does not include attribute data derived from user check-ins.

19 . The method of claim 1 , wherein the received attribute data does not include attribute data derived from user check-ins.

20 . The method of claim 1 , wherein said automatically assigning one or more semantic tags is in response to a user input via a user terminal;
wherein the method further comprises: providing one or more of the predicted semantic tags to the user via the user terminal.

21 . The method of claim 20 , wherein the user input comprises the POI or a search request for a POI.

22 . The method of claim 20 , wherein the user input comprises a proposed semantic tag; and
wherein said providing one or more semantic tags to the user comprises providing additional or alternative semantic tags to the proposed semantic tag.

23 . The method of claim 22 , wherein the user input further comprises the POI or a search request for a POI.

24 . The method of claim 22 , wherein said provided one or more semantic tags comprise one or more tags related to the proposed semantic tag in a hierarchy.

25 . The method of claim 24 , wherein said providing the one or more semantic tags comprises generating for display on the user terminal a visualization of the one or more related semantic tags.

26 . The method of claim 1 , further comprising:
receiving a search request for a POI from a user via a user terminal, the search request comprising a category and a geographic location; in response to said request, searching the database; retrieving one or more POIs having spatial attribute data corresponding to the received geographic location and having one or more of the predicted semantic tags corresponding to the received category; and generating for displaying on the user terminal the retrieved one or more POIs.

27 . The method of claim 26 , wherein the retrieved one or more predicted semantic tags match the received category.

28 . The method of claim 26 , wherein the retrieved one or more predicted semantic tags are related to the received category in a hierarchy of categories stored in the database.

29 . A method for automatically assigning one or more semantic tags to a point-of-interest (POI) using a processor, the POI being represented by attribute data stored in a database, the method comprising:
receiving a request from a user via a user terminal; receiving cultural data for the user; in response to the request, receiving the attribute data for the POI from the database; generating a dataset including data corresponding to a selected set of attributes from the received cultural data and the received attribute data for the POI, the selected set of attributes comprising a POI name for the POI and at least one cultural attribute, wherein the data correspond

Assignees

• Naver Corp

Inventors

• Lagos Nikolaos
• Ait-Mokhtar Salah
• Calapodescu Ioan

Classifications

• G06N3/044

  Recurrent networks, e.g. Hopfield networks · CPC title

• G06N3/048

  Activation functions · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06N3/0499

  Feedforward networks · CPC title

• G06N3/0442

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