Automated Company Background Screening System

Tan, Kai Yuan (2022) Automated Company Background Screening System. Final Year Project (Bachelor), Tunku Abdul Rahman University College.

Text RDS_TanKaiYuan_Fulltext.pdf Restricted to Registered users only Download (1MB)

Abstract

The purpose of this project is to develop an automated system to identify the fraudulent company based on company reviews. The number of businesses is increasing yearly while some of the companies are not registered or authorized by the Companies Commission of Malaysia (SSM) and operating illegally. The development of an automated company profiling system could help the users to avoid from being scammed. The developed system should able to perform web scraping, text preprocessing, fraud detection, and sentiment analysis over the company reviews from the employment website. During the project development, the system is built by using Python, Jupyter Notebook, Microsoft Visual Code and Azure virtual machine for deployment of system as an API. In addition, this project is following CRISP-DM model as standard to produce the system. To perform the topic modeling, Latent Dirichlet Allocation (LDA) is the chosen model to generate topics for performing the fraud detection. The words from the topics will be used to compare with the fraud-related words for fraud probability calculation. Besides, the sentiment analysis is also performed to have an overview at the company sentiment based on the extracted company reviews. A 50 of legitimate company name were tested using the system and all the companies are having 0% of fraud probability as expected as there no exist fraud-related reviews for those companies. Lastly, the system is successfully deployed as API integrated with both company website evaluation function and company review evaluation (company background screening) function. In conclusion, the system is able to perform fraud detection and sentiment analysis over the extracted company reviews on Glassdoor, Indeed and JobStreet.