ABC Corporation needs a system for managing employees, specifically in the area of recruitment and maintaining the records of prospective employees.
The rapid advancement and growth in ICT (information technology systems) in recent years has led to an increasing number of organizations and people using the Internet to seek jobs, as well as for career development. Many forms now make us of online systems of knowledge management for hiring workers exploiting the benefits provided by the world wide web. Such systems are known as e-recruitment systems; they make the process of human resources recruitment automatic. In the past years, e-recruitment systems have been explosive, making it possible for recruiters to target a large pool of potential employees at a small cost (Naim,Tanveer,Gildea, & Hoque, n.d.). However, such a system can be overwhelming to recruiters and HR departments that must pore through applicants and potential employees’ resumes and CV’s to evaluate the suitability of all applicants for a given job position. An efficient approach would be to automate the process of analyzing and evaluating individual CV’s and resumes from sometimes thousands of applicants to arrive at the most suitable candidate for the job (Anrah,& Sokro, 2012). There have been several systems of e-recruitment that have been proposed in the past to help make the process of recruitment in online platforms fast and efficient.
While other researchers have come up with innovative methods for efficient e-recruitment systems, one fast advancing and highly useful technology has not been fully incorporated into e-recruitment; machine learning is a fast developing technology that can greatly improve such applications. The proposed system will make use of machine learning algorithms to enhance the efficiency and accuracy of online recruitment systems. Machine learning entails the use of AI (artificial intelligence) in that gives systems the capability to learn and make improvements in their functioning from experience, without explicit programming. With machine learning, computer programs that can access various forms of data and utilize this data for self learning, based on various algorithms.
Many e-recruitment systems have been proposed, such as E-Gen systems that analyzes and categorizes unstructured job offers (Kessler, Torres-Moreno, & El-Beze, 2007) and analyzing and ranking candidates in terms of relevance. Some approaches such as those used by the Comm-On system makes use of semantic web technologies in HRM, where the personality traits of the candidates are used for recruitment consideration; the traits are filled in by the job applicant. The employees are then matched with possible employers and jobs using relevance feedback recommender systems and IT systems. Semantic matching and analytical hierarchical processing (AHP systems can also be used. The AHP, proposed by Faliagka, Tsakalidis & Tzimas, (2012) investigated past methods of e-recruitment and evaluated the context of modern ICT development and growth and proposed a system in which the resumes of candidates are are modeled in HR-XML and uses the AHP to rank the candidates using an experimental setup with promising results (Shade,Samuel,& Samuel, 2012). Another approach that has been proposed is through the use Neuro Linguistic Processing (NLP) technology based on the General Architecture of Text Engineering (GATE) to formulate the necessary rules for generating annotations when analyzing CV’s (Amdouni & Abdessalem, 2010). The authors sought to develop an NLP based system for automatically representing resumes using standard modeling language.
While the developed and proposed methods are no doubt effective at enhancing e-recruitment goals, they suffer a major setbacks due to the discrepancies inherent in CV and resume formats that are inconsistent, as well as contextual and structural information that is inconsistent. Further, the proposed and developed systems cannot evaluate some secondary characteristics inherent in resumes and CV’s , such as coherence and style; factors that are of great importance in the evaluation of CV’s. Further, there is a dearth of proposed systems that can utilize machine learning algorithms for fast and accurate analysis and ranking of CV’s and linking such analyses with jobs in a cost effective way that can help HR departments and recruiters achieve their recruitment objectives.
The aim of this research is the development of an integrated company-oriented e-recruitment framework that automates the nomination assessment Furthermore pre-screening procedure. Its goal is should ascertain the applicant’s significance scores, which reflect how great their profile fits the positions’ determinations. In this section, we available a review of the suggested framework structural engineering and hopeful positioning plan.
The proposed system will be developed and tested using an experimental research approach where algorithms will be developed and incorporated into the architecture of the system and then tested for functionality and performance. The system is evaluated based on te effectiveness of assessing consistent resume information relevance to rank candidates, and then comparisons made with candidate ranking achieved by human recruiters.
The proposed e-recruitment system in this paper will flatmate the ranking of candidates using a credible criteria that firms can easily integrate into their existing HRM infrastructure. The selection criteria to be used with the system include;
A module for job application- This implements input forms that enable candidates to apply for the job, online. The candidate has an option of logging in with their Linked In profiles credentials from which the system automatically extract the objective criteria for selection directly from the account.
Personality Mining Module- The candidate can have the URL of their blog used in the system; the system then uses linguistic analysis to the candidates blog posts to derive features that reflect the personality of the applicant.
Module for Applicant Grading- This module combines the selection criteria for the candidate to derive a relevance score for the candidate. The grading function is generated through supervised algorithms with the capacity for learning
The qualifications of every applicant and the relevance score of the candidates is stored in the system database. After the recruitment process online, the top candidates are invited to participate in an interview. The system functions in such a way that candidates are not required to enter information manually or participate in personality tests that can be time consuming; this is to ensure the system is user friendly.
With HRM departments easily being overwhelmed with CV’s and job applications because of having to manually evaluate applications (Stone,Lukaszewski,Stone-Romero, & Johnson, 2013), this system will automate and make this process highly efficient and accurate. The manual process is complex, error prone, and can be subject to personality bias and must be adjusted every time the criteria for selection is changed (Stone, & Dulebohn, 2013). The proposed system will leverage machine learning algorithms to build the ranking for candidates automatically, fast, and accurately. Sufficient input training data will be required to train the system using previous decisions made during candidate selection. The ‘learning to rank’ algorithms will be applied in this; the typical process is shown below;
Past applications by candidate are used as the training set and are represented by the vectors xi(k), along with the expert judgment of a recruiter on the candidates relevance score denoted by yi. The features of the candidate can be evaluated numerically on a scale such years working at a job and years of experience; or using a Boolean variable that represents whether the applicant has a specific skill or not in their CV or Linked In profile. The training set is then fed into a learning algorithm that develops a model for ranking to predict the judgment of the recruiter when the candidates vector features is used as the input.
Personality traits are critical in job selection in many cases, a function many e-recruitment systems overlook. Various information found on the web about a person, such as their blog posts, social media comments, or personal web pages are reliable in predicting the personality of a person. Linguistic analyses of blog posts can accurately derive the personality traits of an individual; their emotions and moods can also be derived reliably . The personality traits of the candidates will be derived using Linguistic Inquiry and Word Count (LIWC) that makes use of a dictionary of word stems that are classified in specific systematic and psycho linguistic semantic word categories (Ro?z?ewski & Lange, 2017). The regression model was trained using a recruiters’ scores to predict the extraversion of a candidate based on their LIWC scores categories. A linear regression model was then used to predict the extraversion score, E and equation 1 below developed
Where S is social words frequency from the LIWC, P is the positive emotion works, while N is the negative emotion words frequency.
Here, machine learning techniques are leveraged in solving the problem of candidate ranking in e-recruitment systems. A scoring function h(x) gives the candidates relevance score. The relevance score variable is continuous so the applicant ranking problem can be reduced to a regression problem in the scoring function of the candidate must be learned through supervised learning techniques. The candidate’s relevance degree yi, is derived by the score function h(x) from his feature vector function xi. The xi in this context is made up of m attributes [a1……, am] corresponding to the selection criteria of the candidate. An approximation of the true scoring value is derived from the training set, D. In this system, the training set is made up of a set of N from past candidate scoring
The yi of the ith candidates relevance score is predicted as being the selection criteria of the linear function that is made up of the feature vector of the candidate xi, plus the regression error e ;
One hundred applicants having a LinkedIn profile as well as a blog account were compiled and used for this experiment; the participants were searched from Linked In profiles or their blog posts on Google blog API; the requirement was that the people had a technical background. The technical positions were at a company (not disclosed), where the required job positions were a junior programmer, sales engineer, and senior programmer. The sales engineer needs a person with high extraversion, for junior programmers, the requirement is knowledge of specific languages, while experience is the most desirable attribute for the senior programmer.
The assumption made is that each of the 100 applied for the three jobs requiring a technical background. The applicants were ranked based on a recruiter and the automated ranking by the system. The correlation coefficients were captured and data collected from the experiment as shown below;
|
Correlation coefficient |
LR |
M5 Tree |
SVR |
REP tree |
SVR, PUK |
|
Junior Programmers |
0.80 |
0.87 |
0.86 |
0.85 |
0.81 |
|
Senior Programmer |
0.65 |
0.61 |
0.62 |
0.69 |
0.75 |
|
Sales Engineer |
0.75 |
0.85 |
0.74 |
0.84 |
0.85 |
|
Correlation coefficient |
LR |
M5 Tree |
SVR Poly |
REP Tree |
SVR, PUK |
|
Pearson Correlation Coefficient |
0.67 |
0.67 |
0.31 |
0.66 |
0.65 |
|
Relative Error |
22% |
19% |
51% |
21% |
20.4% |
Weka data mining software (Sondhi, 2017) was used during the first evaluation for learning to rank algorithms with the scores correlation tested on the system output. The actual scores from recruiters were correlated with those of the automated machine learning system using the Pearson correlation coefficient for four machine learning models, namely LR (linear regression), M5’ (M5’ model tree), REP (REP tree Decision), and SVR (Support Vector Regression). Two linear kernels, namely PUK universal kernel and the polynomial kernel. Te best results were generated by the SVR and the tree models with the PUK kernel generated the best results. Linear regression models, by comparison, performed less impressively showing that the selection models cannot be separated in a linear manner. Te system score’s consistency as shown in table I is dependent on the nature of the job positions. The recruiter judgment for the sales engineer is dominated by the extraversion score, which is highly subjective resulting in an overall score that is highly uncertain. However, the correlation coefficient achieved by the system is still impressive, at 0.85. For junior programmer, aspects such as loyalty and education/ knowledge rank high, resulting in a higher correlation coefficient of 0.87. for the senior programmer, experience ranks higher, hence the correlation coefficient of up to 0.75. The personality mining scheme was tested in the second test and a significant positive correlation of 0.67 was obtained.
Conclusions
This paper proposed , implemented, and tested a system for e-recruitment for ABC that makes use of machine learning algorithms. Candidates used were selected using Google Blog API and those with a blog post and Linked In profile selected to participate. The testing was based on regression models and decision trees with automatic score ranking and recruiter ranking. A significant positive correlation of up to 0.87 was found for the accuracy of the scores, showing that the automated system can give as accurate projection as those done by a human. The advantage of the system is that it is fast, accurate, and not prone to subjective bias. The system, based on results, is effective in identifying the extraversion of the candidates and ranking them accurately
References
Essay Writing Service Features
Our Experience
No matter how complex your assignment is, we can find the right professional for your specific task. Contact Essay is an essay writing company that hires only the smartest minds to help you with your projects. Our expertise allows us to provide students with high-quality academic writing, editing & proofreading services.
Free Features
Free revision policy
$10Free bibliography & reference
$8Free title page
$8Free formatting
$8How Our Essay Writing Service Works
First, you will need to complete an order form. It's not difficult but, in case there is anything you find not to be clear, you may always call us so that we can guide you through it. On the order form, you will need to include some basic information concerning your order: subject, topic, number of pages, etc. We also encourage our clients to upload any relevant information or sources that will help.
Complete the order form
Once we have all the information and instructions that we need, we select the most suitable writer for your assignment. While everything seems to be clear, the writer, who has complete knowledge of the subject, may need clarification from you. It is at that point that you would receive a call or email from us.
Writer’s assignment
As soon as the writer has finished, it will be delivered both to the website and to your email address so that you will not miss it. If your deadline is close at hand, we will place a call to you to make sure that you receive the paper on time.
Completing the order and download