A Graph Theoretic Approach for Multi-Objective Budget Constrained Capsule Wardrobe Recommendation

Usage

We make use of three real world e-commerce datasets.

1. Datasets/Amazon-1
2. Datasets/Amazon-2
3. Datasets/Polyvore

Our algorithms are implemeted in C++ (g++ complier with version (Ubuntu 7.5.0-3ubuntu1~18.04) 7.5.0), and data preprocessing has been done in python3 (with version 3.6.9).

To generate graph for each dataset csv, run the script.sh from the respective folder. And to generate real-value edge-weighted graph for each dataset csv, run the script_prob.sh from the respective folder.

For example, in case of Amazon-1 dataset:

    $cd Datasets/Amazon-1
    $bash script.sh
    $bash script_prob.sh

The graphs in required format are stored in input_graph.txt for binary weighted edges, and input_graph_prob.txt for real-value weighted edges.

To run the code, execute the run.sh script

For example, in case of Amazon-1 dataset:

$bash run.sh
Enter Dataset File Path: ./Datasets/Amazon-1/input_graph.txt

Results are stored in result/result.csv

To get the results for real-value edge-weighted graph, execute run_prob.sh script.

For example, in case of Amazon-1 dataset:

$bash run_prob.sh
Enter Dataset File Path: ./Datasets/Amazon-1/input_graph_prob.txt

Results are stored in result/result.csv

Please cite our paper.

    @article {
            PatilBanerjee:2021:MOBCCWR,
            author = {Shubham Patil and Debopriyo Banerjee and Shamik Sural}, 
            title = {A Graph Theoretic Approach for Multi-Objective Budget Constrained Capsule Wardrobe Recommendation}, 
            booktitle = {ACM Transactions on Information Systems}, 
            volume = {1}, 
            number = {1}, 
            pages = {1:1--1:33}, 
            year = {2021} 
    }