Reviewer Finder and Author Prediction

2019 COE Engineering Design Project (EB03)

---

Faculty Lab Coordinator

Ebrahim Bagheri

Topic Category

Software Systems

Preamble

The task of performing authorship attribution in written work is becoming increasingly important given the abundance of user generated content in social platforms and peer-reviewed content. This project will focus on identifying the authors of a piece and likewise appropriate content reviewers when appropriate.

Objective

The objective will be to perform predictive analytics for authorship attribution.

Partial Specifications

+ identify strong features for authorship attribution
+ determine features for appropriate reviewers for published content
+ train predictive models to find appropriate reviewers
+ evaluate the work and compare it against the state of the art

Suggested Approach

his work will build on state of the art literature on the topic of authorship attribution. The students will start by reading the following papers:

+https://www.sciencedirect.com/science/article/pii/S0957417410002083
+https://www.sciencedirect.com/science/article/pii/S0167923612000899

Group Responsibilities

+ Work with gold standard datasets such as DBLP
+ perform extensive feature engineering
+ implement integrated code in Python to automatically extract features
+ use CNN models to extract unsupervised features
+ train classification models for predicting authorship
+ systematically evaluate the work

Student A Responsibilities

+ Work with gold standard datasets based on DBLP
+ perform extensive feature engineering
+ implement integrated code in Python to automatically extract features

Student B Responsibilities

+ Work with gold standard datasets based on DBLP
+ use CNN models to extract unsupervised features
+ train classification models for predicting fake news

Student C Responsibilities

+ Work with gold standard datasets such as DBLP
+ train classification models for predicting authorship

Student D Responsibilities

+ Work with gold standard datasets based on DBLP
+ implement integrated code in Python to automatically extract features
+ train classification models for predicting authorship
+ systematically evaluate the work

Course Co-requisites