# SNIDSA

**Repository Path**: hummingb1rd/snidsa

## Basic Information

- **Project Name**: SNIDSA
- **Description**: No description available
- **Primary Language**: Unknown
- **License**: MIT
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2021-03-19
- **Last Updated**: 2021-03-19

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# Sequential-Neural-Information-Diffusion-Model-with-Structure-Attention

This repository provides reference codes as described in the paper:

>**A Sequential Neural Information Diffusion Model with Structure Attention.**  
>**Zhitao Wang, Chengyao Chen and Wenjie Li.**  
>**CIKM, 2018.** 

## Environment
The code is implemented with Tensorflow. Requirments:  
 1. Python 2.7  
 2. Numpy  
 3. Tensorflow  
 4. tqdm (for training process display)   

## Run
Defalut:  

    python train.py  
    
Or run with optional arguments:  

    python train.py -d data/hc-exp -l 0.001 -x 32 -e 64 -b 32 -n 200 -f 5
Check the arguments as:  

    python train.py -h
    -l, --lr (learning rate)  
    -x, --xdim (embedding dimension)  
    -e, --hdim (hidden dimension)  
    -d, --data (data path)  
    -g, --gpu (gpu id)  
    -b, --bs (batch size)  
    -f, --freq (validation frequency)  
    -n, --nepoch (number of training epochs)

## Data Format
### Cascades:
>u1_id,u2_id,...,ul_id:u1_time,u2_time,...,ul_time (time is optional, we do not use)     
>Example: 334,478,398,222:75.117015,77.968750,78.757250,80.020426     
Or     
>u1_id,u2_id,...,ul_id     
>Example: 334,478,398,222     

### Graph:
>u1_id u2_id weight (weight is optional, we do not use)     
>Example: 0 132 0.010008  
Or     
>u1_id u2_id     
>Example: 0 132     

## Data Creation
You can create graph and generate cascades through the software NETINF (http://snap.stanford.edu/netinf/)
## Citing
    @inproceedings{Wang:2018:SNI:3269206.3269275,
    author = {Wang, Zhitao and Chen, Chengyao and LI, Wenjie},
    title = {A Sequential Neural Information Diffusion Model with Structure Attention},
    booktitle = {Proceedings of the 27th ACM International Conference on Information and Knowledge Management},
    series = {CIKM '18},
    year = {2018},
    location = {Torino, Italy},
    pages = {1795--1798}
    }