FredZhang7 commited on
Commit
07511bd
1 Parent(s): fc35017

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +10 -1
README.md CHANGED
@@ -2,6 +2,15 @@
2
  license: apache-2.0
3
  ---
4
  The plots below show the regression line and correlation coefficients of 20+ features extracted from 593,430 online websites (the features train set) and whether the URL is malicious.
5
- If we could plot the lifespan of URLs, we could see that the oldest website has been online since Nov 7th, 2008, while the most recent phishing websites appeared as early as July 10th, 2023. As we can see, there's no correlation between `is_malicious` and the columns `meta_percentage`, `mouseover_changes`, `right_click_disabled`, and `popup_window_has_text_field` as of 2023, contrary to the [analyses of researchers in 2013 on phishing detection](./Phishing_Websites_Features.docx).
 
 
 
 
 
 
 
 
 
6
 
7
  ![Phish Eater Data Analysis](https://i.imgur.com/ICMudAG.png)
 
2
  license: apache-2.0
3
  ---
4
  The plots below show the regression line and correlation coefficients of 20+ features extracted from 593,430 online websites (the features train set) and whether the URL is malicious.
5
+ If we could plot the lifespan of URLs, we could see that the oldest website has been online since Nov 7th, 2008, while the most recent phishing websites appeared as late as July 10th, 2023.
6
+
7
+ As we can see, there's no correlation between `not_indexed_by_google`, `is_malicious` and the columns `meta_percentage`, `mouseover_changes`, `right_click_disabled`, and `popup_window_has_text_field` as of July, 2023, contrary to some [analyses of researchers in 2013 on phishing detection](./Phishing_Websites_Features.docx).
8
+ The majority of features have very weak correlations with `is_malicious`, while a minority has a weak correlation. Is this problematic for training? Not really.
9
+
10
+ I split the classification task into two stages in anticipation of the limited availability of online phishing websites due to their short lifespan:
11
+ 1. a small multilingual BERT model to output the confidence level of a URL being malicious to model #2, by finetuning on 2,154,175 legitimate and malicious URLs
12
+ 2. (probably) LightGBM to analyze the confidence level, along with roughly 19 extracted features
13
+
14
+ This way, I can make the most out of the limited phishing websites avaliable.
15
 
16
  ![Phish Eater Data Analysis](https://i.imgur.com/ICMudAG.png)