Searching for a gem on backpages : Using Perl, MySQL & VB.NET

[mas0]

This is fairly techny stuff so bear with me or skip to the next post.

After a fair bit of disappointment from BP (And quite a few agencies) and mourning the departure of Dasha@cupids, I wondered if I could potentially improve my odds of finding a gem on BP. And then one day while drinking some redbull at 2am the idea hit me. Why not automate my quest and have an automated system that can filter out the ads and find my next Sp Gem.

I came up with the following premises:

• I have a FreeBSD virtual server
• I have perl & mysql installed (and not really doing much on this box) and some knowledge on how to do a few things
• I have the logic
• I've got some time to waste
• I'm bored

So I decided hey what the hell let me start on a little project during my spare time and see how it goes. I came up with a list of requirements and things that I could automate, the plan was to extract an Ad from BP and perform some logic and analysis on it. The basic draft of my requirements was to create a Robot that can

• Grab all the current ads automatically.
• Extract phone numbers
• Extract Pictures
• Guess the ethnicity based on keywords (I am partial to Brown and white girls)
• Find and filter out the racist ones (I'm not caucasian)
• Filter out spam (Agency and B&S - semi automatically)
• Keep a record of who I receive text messages from and "silence" their ads
• Give me an ability to add SPs to lists and block their ads
• Find and filter SPs by location (I don't enjoy traveling far for hobbying)
• Be able to search for similar Ads by phone number
• Include shortcuts to my filtering for googling numbers and using tinyeye to find fakes\duplicates

• Grab all the current ads automatically.
• Extract phone numbers, Ad titles, descriptions and such
• Extract Pictures
• Guess the ethnicity based on keywords (I am partial to brown and white girls)
• Find and filter out the racist (or racially selective) ones (I'm not caucasian)
• Filter out spam (Agency and B&S - semi automatically)
• Keep a record of who I receive text messages from and "silence" their ads
• Give me an ability to add SPs to lists and block their ads
• Find and filter SPs by location (I don't enjoy traveling far for hobbying)
• Be able to search for similar Ads by phone number
• Include shortcuts to my filtering for googling numbers and using tinyeye to find fakes\duplicates

I started my little project about a week ago and so far I am extremely satisfied with the back end;

I have accomplished almost everything and I have to say I am quite content, I have been able to store and filter out most of what I want as shown in the following screenshots. I have been successful in eliminating a lot of B&S, Agency scams etc. I am presented regularly with a fresh list of Ads which I am currently (for the time being) filtering manually with sql queries. My next step is to create a VB.NET front end... which unfortunately I am quite a newbie to, so it's taking me a LONG time to actually build a decent interface for. But overall for about 10-15 hours worth of work... I'm not unhappy.


Structure of my DB



With Phone extraction, location extraction has been a massive pain in the ass



If anyone has any suggestions or ideas I could use please feel free to let me know

 

[George The Curious]

You have way too much time dude. But look into openCV image feature recognition to scan the pictures for your liking or and possible fakes.

 

[mas0]

Come on. I'd actually like to be able to choose myself man. I can't make the machine pick my next date ... that would be too... LOL

 

[George The Curious]

Come on. I'd actually like to be able to choose myself man. I can't make the machine pick my next date ... that would be too... LOL

if you can make a program to categorize by race (white, black, asians, other), and type (tall, petite, bbw , milf, older) by scanning pictures?

 

[mas0]

Actually yeah I could most likely do that, BUT I use simple keywords in the ads to determine that. it has a high percentage of success of course with some misses but it's good enough I suppose.

 

[Steamdelux]

mas0, do you know node.js? Wanna take this a step further?

 

[Steamdelux]

if you can make a program to categorize by race (white, black, asians, other), and type (tall, petite, bbw , milf, older) by scanning pictures?

How about a crawling website that allows people to tag posts, utilize technology to create strong associations between multiple similar posts and, more interestingly, the ability to let people vet

a hooker facebook, who is down for a hackaton?

 

[mas0]

mas0, do you know node.js? Wanna take this a step further?

Not familiar with node.js at all. I was going to attempt working with that for a home automation project using beaglebone but... life got in the way and I2C coding made c++ a better option.

How about a crawling website that allows people to tag posts, utilize technology to create strong associations between multiple similar posts and, more interestingly, the ability to let people vet

a hooker facebook, who is down for a hackaton?

As you can see I am already tracking pretty much every change there is. I perform some searches manually but automating that too is absolutely feasible. The main issue with your idea is the cost-benefit. Right now I am running my own server and it works great for _me_. I have even incorporated a system where my text messages are saved in the DB so I can filter out ladies I've seen before or eventually even use keywords to categorize further, so it's not readily portable. To add to this, most of the BP girls are terrible with spelling and entering locations properly... again refer to the image above, it makes GeoLocation even harder to automate, it's a LOT of work to clean this up. But I'm curious to hear your ideas.