Pandemic-Aware Local Recommendation

Code Details

"sample_LP.py": Sample code for the proposed linear programming approach

INPUTS: All the details required for one instance of ranking (***note (lambda_1+lambda_2) <= 1)
    E (array) = current exposure of businesses in an array
    capacity (array) = safe capacities of businesses in an array
    relevances (array) = rele
    k (int) = size of the recommendation
    beta_fraction (float) = beta/beta_max (this ultimately decides how much of exposure we are trying to guarantee for every business)
    lambda_1 (float) = weight (fraction) for the sustainability term in the objective
    lambda_2 (float) = weight (fraction) for the safety term in the objective
    
OUTPUT: 
    reco = A k-ranked list of businesses
    E = updated exposure of businesses

Dataset Information

List of files

Customer_Choice_Survey.csv
NYC_Google.csv
NYC_Yelp.csv
SF_Google.csv
SF_Yelp.csv

Field Details in Each File

• "Customer_Choice_Survey.csv": Each respondent was first asked some basic details. Then 7 rounds of ranking questions were asked. In each round, they were given a list of 10 restaurants with random combinations of rating, distance and cuisine. They were asked to rank top 5 one-by-one out of those 10 provided. This becomes evident from the question titles provided the file.
• "NYC_Google.csv" and "SF_Google.csv": Local recommendations received on Google Local (Google Maps) for different customer locations in New York and San Francisco.

"customer_location": location of the customer where she gets recommendation
"rank": rank of the restaurant in the recommended list
"id": restaurant's id internal to google
"latitude": latitude of restaurant's geographic coordinates
"longitude": longitude of restaurant's geographic coordinates
"name": name of the resturant
"price_level": cheap/costly level
"rating": average rating of the restaurant
"rating_count": number of ratings collected for the restaurant
"address": address of the restaurant

• "NYC_Yelp.csv" and "SF_Yelp.csv": Local recommendations received on Yelp for different customer locations in New York and San Francisco.

"customer_location": location of the customer where she gets recommendation
"rank": rank of the restaurant in the recommended list
"id": restaurant's id internal to yelp
"latitude": latitude of restaurant's geographic coordinates
"longitude": longitude of restaurant's geographic coordinates
"name": name of the resturant
"rating": average rating of the restaurant
"rating_count": number of ratings collected for the restaurant
"address": address of the restaurant
"url": link to the restaurant's yelp page

Citation Information

Please cite the following paper if you use this dataset or the code.

• "Towards Sustainability and Safety: Designing Local Recommendations for Post-pandemic World"
  Gourab K Patro, Abhijnan Chakraborty, Ashmi Banerjee, Niloy Ganguly.
  In proceedings of Fourteenth ACM Conference on Recommender Systems (RecSys-2020), Virtual Event, Brazil.
  

Or you can also use the following bibtex.

@inproceedings{10.1145/3383313.3412251,
author = {Patro, Gourab K and Chakraborty, Abhijnan and Banerjee, Ashmi and Ganguly, Niloy},
title = {Towards Safety and Sustainability: Designing Local Recommendations for Post-Pandemic World},
year = {2020},
isbn = {9781450375832},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
url = {https://doi.org/10.1145/3383313.3412251},
doi = {10.1145/3383313.3412251},
booktitle = {Fourteenth ACM Conference on Recommender Systems},
pages = {358–367},
numpages = {10},
keywords = {COVID-19, Local Recommendation, Google Local, Yelp, Safety, Social Distancing, Sustainability, Bipartite Matching},
location = {Virtual Event, Brazil},
series = {RecSys '20}
}

Extended Version of The Paper

Towards_Safety_and_Sustainability_Extended.pdf

You can also access the extended version of the paper here.