Skip to content

detections

G edited this page Jan 8, 2021 · 8 revisions

Detections model

This project utilizes a custom multi layer detection model - The more detections you have for a specific target, the better detection rate and results

social-analyzer/sites.json file has a list of items called website entries

[
  {
    "url": "This is the target website with {username}",
    "detections": [
      {
        "return": "What does the evaluation from finding some string return? (false or true)",
        "string": "A string to detect on",
        "type": "There 3 type of detections: ocr, normal or advanced"
      }
    ],
    "selected": "Auto select this item on loading the project? (false or true)",
    "timeout": "Add Time out (in seconds), if 0 then it will use 5 seconds",
    "implicit": "Add Implicit Time out (in seconds), if 0 then it will use 5 seconds",
    "type": "What is the website category? (One word E.g. Fun)"
  }
]

The detections key is a list of items, each item is a detection that returns true or false when the logic is evaluated and matched the return value

"detections": [
    {
      "return": "false",
      "string": "page not found",
      "type": ""
    }
]

Writing Detections

Example: If a website does not have page not found then add 1 to the counter. If the website has your name is: then add 1 to the counter. This means that we have 2 out 2 detections that are valid in the logic

"detections": [
    {
      "return": "false",
      "string": "page not found",
      "type": "normal"
    },
    {
      "return": "true",
      "string": "your name is:",
      "type": "normal"
    }
  ]

Example: The target website https://example.ccc has a publicly available profile for johndoe that can be access at https://example.ccc/p/johndoe. The result is johndoe profile that does not include page not found and has your name is: johndoe

[
  {
    "url": "https://example.ccc/p/{username}",
    "detections": [
      {
        "return": "false",
        "string": "page not found",
        "type": "normal"
      },
      {
        "return": "true",
        "string": "your name is:",
        "type": "normal"
      }
    ],
    "selected": "false",
    "timeout": 0,
    "implicit": 0,
    "type": "other"
  }
]

You can add {username} to the second detection item for more accuracy as the following

[
  {
    "url": "https://example.ccc/p/{username}",
    "detections": [
      {
        "return": "false",
        "string": "page not found",
        "type": "normal"
      },
      {
        "return": "true",
        "string": "your name is: {username}",
        "type": "normal"
      }
    ],
    "selected": "false",
    "timeout": 0,
    "implicit": 0,
    "type": "other"
  }
]

Types of Detections:

An Optical Character Recognition (OCR) checks if a website screenshot contains the detection string or not

  • Example: A website returns an image that contains 404 page not found!
  • Detection: Check if page not found in the output from OCR or not
{
  "return": "false",
  "string": "page not found",
  "type": "ocr"
}

A normal detection checks if a website source code contains the detection string or not (It's similar to going to a website and clicking view page source)

You can also detect on some source code like tags etc..

  • Example: A website returns username is: johndoe
  • Detection: Check if username is in the source code or not
{
  "return": "true",
  "string": "username is",
  "type": "normal"
}

An advanced detection checks if the website contains the detection string or not (It's similar to going to a website and right-clicking on an item then choosing inspect feature - This option is very helpful if the target website manipulates DOM)

You cannot detect on source code like tags, I will add this feature later on

  • Example: A website returns `username is: johndoe
  • Detection: Check if username is in the inspected website or not
{
  "return": "true",
  "string": "username is",
  "type": "advanced"
}

Testing Detections

Detection are tested with my Dedicated Testing System called Macaw before and after pushing them to this GitHub (Some tests are Site Status, Name Checking, Similarity Checking & Alexa Metrics)

Contribution

  • Make sure that you have a legit website and has higher Alexa ranking otherwise Macaw will flag it
  • Focus on adding more detections to a website (Do not focus on adding more websites with lower detection quality)
Clone this wiki locally