in order to make money with this idea, you can do the following:
create thinner splitable wireless keyboard
create typing traning software with good interface
better computer stand so you can insert keyboard closer and introduce more ports
sell things together, introduce many usages to let people see what really matters
send ads along with your media, not just fun videos and news
OSINT/recon 其实就是社工 但是一般人喜欢把社工库和社工分开 因为社工库是社工收集来的数据集合 而社工则是一个过程
loading/transforming leaked txt files will be time-consuming. use pypy to speedup the process. use database specific batch processing method to import the data.
entity fragmentation in followthmoney is kind of for “entity recognition in multiple social platforms”, suitable for finding patterns/clients in large leaked databases.
socialscan Python library for accurately querying username and email usage on online platforms
Zen collect email on github
maigret a powerful fork of sherlock (customizable, finding accounts by username, but only having 300+ sites ) with 2000+ sites
emailAll collect email by passing domain
emailfinder find email by domain
theHarvestor email, subdomain and names collector
gitscan scan for email and password (if possible) with predefined domains and rules by searching github
ghunt needs companion browser plugin to get credentials. can collect info on given email
EMAGNET collect database leaks, email and password from recent pastebin records
occrp (anti corruption & crime) aleph is a bad source for getting email (anonymous/unauthorized user can only get hundreds, having no clue what the email relates to). however, it has a tool called follow the money which works with csv files and exports cypher to neo4j
search for leaked database on github
linkedin database leak 2021 (hate mega since it has download quota)
you typically find links to these databases on anonfiles.com (or else), so query like site:anonfiles.com email rar in duckduckgo (no DMCA censorship)
breachedforum’s index contains “credits only” threads which requires 4-8 credits to unlock. to get credits you need to create thread (which will earn 0 credit) and get 1 credit per reply. post to trivial threads like manga.
in leakbase you earn credits and download leaked databases easier. it has official telegram bot claims to leak free databases everyday.
find telegram bots collection in privacy.club (only OSINT bots) and here (with many other bots)
摆烂bot 永久免费
(gone?) FreeSGKbot
although i find many leaked databases as torrent, but those torrent search engines usually collect video/movies instead of anything related to leaked database.
44.65GB QQ 微博社工库 or qq8e/qq 基本上传的就是这个了 或许可以在QQ群里面找到一些别的社工库 使用这个数据库的还有 q绑 (有反调试 带去水印工具 但是其实这个到处都有吧) aiuys’s retrofit 后台是privacy (只是部分的可以导入 其他的自行处理)
Using aMule on macOS, Kad is firewalled (2.2.1 works well said by people, but I’d not use macOS), reason unclear. Maybe on Linux or Windows it will be different.
Some (dead) links of other databases in ed2k emule format:
1 | ed2k://|file|2010.06-江西移动全库-408万-access.7z|1329999527|5231E1EC5EE1123C6E694AD6399F9807|h=DORZC7XFNG63ZWX6C3RSN3Q7CWZXG5G4|/ |
reg007 counterpart only check if that account “exists”, but no actual account shown. search for reg007 on github you will get a bunch of links relating to OSINT.
holehe only check if an email address is registered as account elsewhere using “forget my password” APIs.
sreg is found from a collection of security tools, which is a deprecated tool for getting registration status with phone/email/username on multiple chinese platforms.
Usually it only verify existance of given email, like emailhippo (100 requests free per day per ip), or mailforguess checking “gmail”,”laposte”,”protonmail”,”yahoo” emails
Some of them verify email with password: verify email address and password with API of my.com
h8mail email osint tool, can chase and link to social profile, with many API-key required free services
mosint automated email osint tool, requires many API keys.
ignorant checks if a phone number is used for snapchat/instagram
proton mail unofficial client in python
before it was using 2captcha, now using 5sim to receive sms
only login such account through pop3 and imap to prevent revocation
using 2captcha and anycaptcha for captcha solving (paid)
using noptcha or nopecha browser extension (free for 100 captcha solves per day) solving hcaptcha, recaptcha. this extension cannot be used with proxy.
I place the list under /root/.muumuu_emails
the muumuu mail is programmatically connectable using account and password. seems it is using default ports for these services. POP3 is for both sending and receiving. IMAP is for receiving. SMTP is for sending.
this guy’s code is full of hacks. seems only being able to run on his own computer and will break on slightest errors.
he stored potential password combinations and also registered accounts (need testing, some may not work) on this google doc. you can download the sheet named “Sheet1” by this api, which adds double quotes and takes more space than exported from web interfaces, method described here.
tempumail get free temp edu email
email proxy for resending email to you, which I used for github registration (but with a very high block rate without proxy)
icloud’s “hide my email” service seems only provide few email aliases. but according to 3rd party icloud alias generator (cannot be used for chinese version of icloud) you can generate at least 10 aliases. or use hidemyemail-api to login with pyicloud and get aliases as API service. account registered from web without logged in any apple device (maybe virtualbox -> macos has a shot?) will not have email service.
to send email from alias, you can try setting “FROM” address as your alias via smtp protocol, but the credential shall stay the same. the working approach could be platform specific
yahoo provides the most email alias up to 500, but 10 for send only emails. however to get one yahoo account one needs offshore phone numbers.
searching for “site:pastebin.com @yahoo.com” to get some email addresses, also searching in github might help as well.
mailcat find email address by nickname (check if deliverable?)
use other’s links/contents to increase diversity and increase anonymity. put your related contents among them.
email marketing is quantity over quality. know your customers’ preferences and behaviors (language, country, life schedule (by year? month? week? time in a day?)) by linking their accounts on other platforms, telemetry.
email bulk senders are equipped with email templates, statistics (like opened or not, click data monitoring)
vary your email style and content unless you want to get blocked/trashed by servers
premail is an easy-to-use component-based build system for MJML, the email templating language
发送短信 邮件 第一步是收集目标的邮箱和手机号码 收集目标ID 可从社工库中获取 可以根据社工的metadata决定推送内容类别
smsboom by whalefall 远程获取的api 不知道是干啥的 有待研究
这个的原理是收集了大量发验证码到目标的手机号的api
repo转移到了openethan下面
考虑破解这些网站 获取它们的短信验证所需要的credential
https://github.com/WhaleFell/SMSBoom
send sms 1 per day per ip (you might use tor to do ip switching):
https://github.com/HACK3RY2J/Anon-SMS
https://github.com/typpo/textbelt
freesms: (don’t work for my phone number as recipient though, but I found some interesting projects on github relating to free SMS sending, some using OCR to crack captcha and access API)
https://www.afreesms.com/freesms/
yoy might want phone number lists for sending ads via free sms providers
shows [number]@139.com is the only email2sms gateway in china
list of email2sms providers:
https://www.howtogeek.com/howto/27051/use-email-to-send-text-messages-sms-to-mobile-phones-for-free/
http://www.mutube.com/projects/open-email-to-sms/gateway-list/
http://en.wikipedia.org/wiki/List_of_carriers_providing_Email_or_Web_to_SMS
5sim paid sms receive service worldwide
sms auto regist is written in go, utilizing yunjiema.top for sms receiving
online sms receivers are not so reliable (not even usable for yahoo registration), and those found from google searches (like free receive sms (这个网站有反js调试 打开debugger自动暂停执行), which has simple interface for fetching data, and you can search this site on github to get more sources and potential API adaptors like disposable phonebook) have chances to get registered yahoo accounts.
account generator helper including:
1 | Temp email services |
Proxy-Cheap pay by data amount
字节跳动 抖音推荐算法 Wide&Deep
1、muricoca/crab
https://github.com/muricoca/crab
2、ibayer/fastFM
https://github.com/ibayer/fastFM
3、Mendeley/mrec
https://github.com/mendeley/mrec
4、MrChrisJohnson/logistic-mf
https://github.com/MrChrisJohnson/logistic-mf
5、jadianes/winerama-recommender-tutorial
https://github.com/jadianes/winerama-recommender-tutorial
6、ocelma/python-recsys
https://github.com/ocelma/python-recsys
7、benfred/implicit
https://github.com/benfred/implicit
8、lyst/lightfm
https://github.com/lyst/lightfm
9、python-recsys/crab
https://github.com/python-recsys/crab
10、NicolasHug/Surprise
https://github.com/NicolasHug/Surprise
linkedin gdmix simple and memory effective personalized ranking
datawhale fun-rec 推荐系统入门教程
datawhale rechub
image to text, text to image, clip as image/text embeddings
deep recommendation using tensorflow 1.15
不同的人有不同喜好
不同的人和不同的人说话
不同的产品有不同的特征
不同的产品和不同的产品被一起推荐
人对产品的接受度
youzan has an ai platform called trexpark, offering chinese NLP and image models pretrained from e-commerce databases.
https://github.com/youzanai/trexpark
session based recommendation system:
https://github.com/CRIPAC-DIG/SR-GNN
decide the feedback embeddings:
https://huggingface.co/youzanai/bert-product-comment-chinese
conversational embeddings:
https://huggingface.co/youzanai/bert-customer-message-chinese
neo4j developer build a recommendation engine:
https://neo4j.com/developer/cypher/guide-build-a-recommendation-engine/
torch_geometric(PyG) documentation:
setup GCN using PyG:
https://zhuanlan.zhihu.com/p/400078504
tagspace text classification via hashtags:
https://paddlerec.readthedocs.io/en/latest/models/contentunderstanding/tagspace.html
gnn is based on basic data/label models and provide high-level reasoning and predictions.
neo4j graph academy practical usage:
https://graphacademy.neo4j.com/categories/
https://neo4j.com/graphacademy/training-iga-40/12-iga-40-ingredient-analysis/
video segments have different features and orders. predict missing links. predict categories semi-supervised or unsupervised.
video-image-text-music correlation and predict internal relationships, categories.
recommendation system:
paddlerec(multimodal), torchrec(cuda==11.3, build failed due to unable to find ATen from torch/include.)
https://neo4j.com/docs/graph-data-science/current/end-to-end-examples/fastrp-knn-example/
link prediction:
https://github.com/Orbifold/pyg-link-prediction/blob/main/run.py
how to use pyg for link prediction:
https://github.com/pyg-team/pytorch_geometric/issues/634
dgl, install from source, with link prediction:
https://docs.dgl.ai/tutorials/blitz/4_link_predict.html
https://docs.dgl.ai/guide_cn/training-link.html#guide-cn-training-link-prediction
gnn intro:
https://cnvrg.io/graph-neural-networks/
gnn applications:
Node classification: The objective here is to predict the labels of nodes by considering the labels of their neighbors.
Link prediction: In this case, the goal is to predict the relationship between various entities in a graph. This can for example be applied in prediction connections for social networks.
Graph clustering: This involves dividing the nodes of a graph into clusters. The partitioning can be done based on edge weights or edge distances or by considering the graphs as objects and grouping similar objects together.
Graph classification: This entails classifying a graph into a category. This can be applied in social network analysis and categorizing documents in natural language processing. Other applications in NLP include text classification, extracting semantic relationships between texts, and sequence labeling.
Computer vision: In the computer vision world, GNNs can be used to generate regions of interest for object detection. They can also be used in image classification whereby a scene graph is generated. The scene generation model then identifies objects in the image and the semantic relationship between them. Other applications in this field include interaction detection and region classification.