###### tags: `Wish` # Image Reuse *Joanna Jiang 2019-06-25* --- ## Summary From the Google sheet recording ==850== reused images in ==2190== commerce tikect links, there are ==1502== distinct tikect ids(there are cases that multipel pictures in a single ticket, with [==949==](https://console.treasuredata.com/app/queries/editor?queryId=1210780) distinct merchants, and [==897==](https://console.treasuredata.com/app/queries/editor?queryId=1209714) distinct users involved. Click to see the [ticket_extraction script](http://localhost:8888/notebooks/Desktop/Wish/0625/extract%20ticket_id.ipynb) and [result](https://docs.google.com/spreadsheets/d/1_giPE820Tf4a3NMkApSEb9Pm1Qby-XaEpW3DUmg_o-A/edit#gid=1894037217). Among the 949 merchants, their history refund ratio ranges from 10.9% to 42.7%. And among the 897 users, The refund ratio can get as high as close to 100%. In respons to 3 possible explanations for this image reuse phenomenon, it shows evidence for self-buying and refund abuse. However for the competitor attack one, since the data could look very similar to self-buying case, it can be a possible one and needs more investigation. ## Assumption 1 :::info Self-buying Merchant Trying to Earn Money from Wish Refund. ::: By looking at the billing names of those refunded merchant, I look for evidence of self-buying. It shows some merchants do have self-buying behavior. * Figure 1 shows the merchant with most count of a same billing name as "建新 戴". [See Query 1](https://console.treasuredata.com/app/queries/editor?queryId=1226933) *<div style="text-align: center" > Figure 1 </div>* ![](https://i.imgur.com/XswiwB8.png) * Figure 2 shows the account balance of this merchant. *<div style="text-align: center"> Figure 2 </div>* ![](https://i.imgur.com/zd3MAuQ.png) * Figure 3 shows this merchant's account balance and refund activities. *<div style="text-align: center"> Figure 3 </div>* ![](https://i.imgur.com/OnICmS0.png) We can also validate this assumption by looking at refund records by user. * Figure 4 shows that 2 billing names are linked to a same user_id. [See query 2](https://console.treasuredata.com/app/queries/editor?queryId=1226933) *<div style="text-align: center"> Figure 4</div>* ![](https://i.imgur.com/lCQ8eUD.png) A reasonale suspicion is that this merchant is asssociated with these 2 person to conduct self-buying for Wish Refunds. Usually, they will apply for refund with reasons like "Item Does Not Match Listing" or "Item Does Not Work As Described" In [query 2](https://console.treasuredata.com/app/queries/editor?queryId=1226933), We get a name list of refund fraud users: "清锋 谢", "建新 戴", "启年 戴", "少华 戴", "JIAN XIN DAI", "Qingfeng Xie", "志清 郑", "晓波 郑" etc. It's also true that the same refund fraud user can be linked with several merchants. ## Assumption 2 :::info Refund Abuse Users Trying to Earn Money ::: For this assumption, I find users who are associated with large number of merchants. * Figure 5 shows the user who applied for refund from most number of merchants. *<div style="text-align: center"> Figure 5 </div>* ![](https://i.imgur.com/87GB1xp.png) Looking at his past transactions in Wish Admin - Commerce Transaction, Figure 6 shows the user had a lot of different refund records from different merchants. *<div style="text-align: center"> Figure 6 </div>* ![](https://i.imgur.com/16SERkL.png) It's also helpful to look at the user's shipping address and IP address. Most of the IP address "139.216.17.51" matches with shipping address in Australia. And the shipping names refer to the same person "Dani Eastman". Further more, this user have been using all kinds of reasons to ask for refund. Figure 7 and Figure 8 show some examples. *<div style="text-align: center"> Figure 7: Item did not meet expectations </div>* ![](https://i.imgur.com/LkgbADx.png) *<div style="text-align: center"> Figure 8: Item does not fit </div>* ![](https://i.imgur.com/eN0qpDE.png) This user is very likely to exploit the refund right and repeate order a same product for quantities. Thus it could cause image reuse issue in this case. I conclude that this user is a refund abuse user who tries to simply earn money from refunds. ## Appendix Related Queries: * [image_reuse_m_u_id](https://console.treasuredata.com/app/queries/editor?queryId=1202205) * [image_reuse_merchant](https://console.treasuredata.com/app/queries/editor?queryId=1202304) * [image_reuse_user_full_joins](https://console.treasuredata.com/app/queries/editor?queryId=1208064) * [image_reuse_user_gmv_aov](https://console.treasuredata.com/app/queries/editor?queryId=1202831) * [image_reuse_user_name_country](https://console.treasuredata.com/app/queries/editor?queryId=1208060) * [image_reuse_m_u_name](https://console.treasuredata.com/app/queries/editor?queryId=1226933)