# gpt_killer

## Proposal
- papers given document description

### intoduction

### methodology

WAN 工作原理

简单的说，WAN 是把文本转换成了网络。

对于每一个作者，都会有一个网络

classification原理：对于一个未知文本，首先计算WAN，与已知作者比较，找出最相似的那个

### idea from zhang chi

用新的办法构建word graph,不用WAN 而是其他办法比如

## Questions field

数据维度不统一，如何处理？
## maybe we can use some ways to reduce dimensionality such as edgenet filter to increase sparasity of matrix and keep important nodes

224 function words(maybe we can adjust the num of function words )


but for data only 211


need to construct WAN for chatgpt, but unknown data. only given network.