exascaleinfolab/lbsn2vec

Code release for LBSN2Vec

lbsn2vec's Issues

how to get cities data from raw_checkins.txt

Great job! Lots of amazing and concrete works on your personal page.
however, I download your raw data from your HOME page and I found there is no city and poi mapping in your check-ins and poi data. If there is a preprocessing script to get dataset_connected_{city}.mat?
Thanks in advance.

Test dataset leakage?

Can you explain how the dataset is used? As I see in the dataset, there is an overlap between 'frienship_old' and 'friendship_new'. Therefore, maybe the results of the paper may be wrong.

some mistakes in the data.

Hello, I found some mistakes in the data.
I use the mat file provided here and the full sign-in file for data processing. In the mat file, one item is selected_venue_IDs(venue_index) => venue IDs

However, I found that only the venue id in NYC is correct, and the venue written in other files does not belong to this city.
For example, the first venue id in TKY.mat is 4b11a2c7f964a5200e8123e3,
But by querying raw_poi.txt and api of foursquare, this poi belongs to Tucson, United States.
it does not belong to Tokyo, Japan,
This type of POI appears in the city mat file but does not belong to the city also occurs in other mat files, such as SaoPaulo.mat
I really hope you can explain it here,
thank you very much

Learn

It seems there is no evaluation in the code

I have tried to run the project as the instructions in your code. But after all, the code just printed out some statistics about the data and parameters but not the results or accuracy.

How can I know if the embeddings that the program saved is good or not?