FST Graduate Research Topic


Research Topic: Authorship Forensics - A WebView app on Android
Suggested By: Maiga Chang
Last Update: March 19th, 2025

===============================


 


Route: This research topic can be taken by essay-route or project-route student. 


 


NOTE: for project-route students, a simple usability analysis will need to be done to assess how users perceived the usability of the plugin with a small group of test users (12 to 15 users). 


 


===============================


 


In this evolving landscape of text generation, distinguishing between human-written and ChatGPT-generated content has become increasingly important. Our research group has analyzed and developed a novel approach to authorship attribution, leveraging both Statistical Natural Language Processing (SNLP) and Convolutional Neural Networks (CNN) techniques to differentiate between documents written by humans and ChatGPTs. The research uses 212 abstracts of academic papers written and published by the research group as the human-written set and asks both ChatGPT 3.5 and 4 to generate correspondent abstracts based on paper titles as AI-written set. Models are trained to classify human and AI-written abstract texts in 2-class (i.e., human and ChatGPT) and 3-class (i.e., human, ChatGPT 3.5, and ChatGPT 4) based on their part-of-speech tag frequency distribution patterns. The 2-class model is well-trained in less than ONE minute (i.e., 56.82 seconds) and the 3-class model is well-trained in 7 minutes and 26.076 seconds. The results demonstrate a significant ability of the models to distinguish between human and AI-written text, with precision 0.9682 (F0.5 score 0.95) for the 2-class (human and ChatGPT) testing subset and precision 0.9806 (F0.5 score 0.96) in the 3-class (human, ChatGPT 3.5, and ChatGPT 4) testing subset. The proposed 3-stage Authorship Forensics approach has also been implemented as an open access web application to allow teachers and users to either training their own models or using the existing trained model to get some advice on how the model considers a piece of given text is written by human or AI.


 


The results show that the computer can be trained quickly with acceptable amount of data like existing research use and the trained model can effectively differentiate texts from an unseen subset in terms of human or ChatGPT-written. Since the trained model’s performances have outperformed almost all existing research, this research aims to design and develop 


 


(A) the correspondent functions that 


 




  1. assigns license key for a registered user as well as when the user registers an account at Authorship Forensic Portal;



  2. has a separate PHP that provides a list of available public models as well as the models belong to a license key (if provided) in JSON;



  3. has a separate PHP that receives the recognition job request package in JSON format;   


    1. interprets the request package;   



    2. checks the content integrity;   



    3. validates the license key included in the package;  



    4. provides Authorship Forensic the text and chosen model to do the recognition; and,   



    5. prepares the result package in JSON format and return it to the client.





  4. a set of separate PHPs that


    1. at least works on Brave, Chrome, Edge, Firefox, and Safari;



    2. allows user to enter and save the license key into local storage later to include in the recognition job request package as well as save it into cookies so Authorship Forensic service later can check against the received license key;



    3. retrieve available models for user to choose one as default and save it into local storage for including in the recognition job request package;



    4. allows users to paste a text copied and click to initiate a recognition job;



    5. saves recognition job details like the selected text, check date/time, chosen model, recognition result (when later received), etc. into local storage;



    6. prepares and send the request package in JSON format, including license key, text, model, etc.;



    7. provides users a status list of the reconigition jobs and allows them to filter and shorten the list by choosing dates or date range, entering words (against to links and text) in free text, etc.;



    8. shows users a particular job's details like the text with the received recognition result when they click a job whose status is "complete".





 


B) a Android app that 


 




  1. use WebView to access the entry point of Authorship Forensic service provider;



  2. logging function and log viewer;



  3. and looking for thoughts and plans on more functions that can enhance user's experiences.