FAKE JOB DETECTION SYSTEM IN INDONESIA

• Main Author: Nindyati, Okti
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/54175
• Institution: Institut Teknologi Bandung

In order to produce a reliable and qualified workforce, Indonesian government should utilize all available opportunities to create effective employment solutions. One solution that can be done is to design a qualified training program so that the competence of job seekers can be in line with the industrial needs. To create an optimal training program, Training Need Analysis (TNA) is needed before establish a training program. Identification of job vacancies is the first part of steps that have .to be taken in the TNA. The flow of TNA activities is to conduct the survey request to the company and collect some information about what type of job requirement is required. By using result from the interview, the officer identifies industry needs and makes a report that used as a consideration tool for preparing the future training program. It can’t be denied that existing business processes are still less effective because it cost much time and expense. Therefor it is necessary to apply technology that may help to identify job vacancy more quickly and efficiently. The comprehensive data of job requirement and specification can be identified by collecting job advertisement information from job vacancy website or Applicant Training System (ATS). By using this data, the TNA business process can be more efficient because it gets comprehensive results immediately with reasonable cost. Unfortunately, collecting data with this technique has the disadvantage of requiring steps to ensure that the data is valid because there have been many fraudulent job ads by individuals on behalf of the company. This fraudulent job ads are usually called an employment scam and included as one type of Online Recruitment Fraud. When the job vacancy information contains fraudulent job ads, then the results of the analysis produced by the TNA system do not reflect actual industry needs. In order to avoid invalid analysis, therefore employment scam detection is needed. Employment scam detection is a determination of whether a job vacancy is a fake or genuine job by carrying out the classification process. To reflect the industrial needs in Indonesia, the data used is limited to job vacancies taken from websites or ATS in Indonesia. Research related to the detection of fake job openings already exists but cannot overcome the existing problems. Previous research built a dataset based on an ATS. To get a comprehensive picture of the needs of the workforce, job advertisements must also be taken from various sources in Indonesia. Therefor in this study, researchers created and used a dataset of job vacancies in Indonesia based on job advertisements on Google Jobs that contained various job vacancies in Indonesia. Using Google Jobs can also make the proposed job scam detection feature more general and can be used widely in other case studies in different countries. The use of methods and features in research related to the detection of fake job vacancies that currently exist are not suitable for use in the fake job detection in Indonesia. This is evidenced by the low average value of accuracy, precision and recall by using several binary calcification algorithms. Therefor it is necessary to formulate a framework for job vacancy detection in Indonesia. The different characteristics of fake job vacancies in Indonesia and data collection from various types of service providers bring new features that can be considered in the fake job detection framework. The indonesian fake job detection framework adds features related to service provider information and adds contextual features namely behavioral and credibility features related to company and job vacancy information. This framework testing result indicate that it can increase the value of accuracy, precision and recall of various algorithms used. This proves that this framework can be used to detect fake job in Indonesia and may complement the shortcomings of previous research.