Download This PaperA Performance Comparisons of Machine Learning Classification Techniques for Job Titles Using Job Descriptions
8 Pages Posted: 1 May 2020 Last revised: 28 May 2020
Date Written: May 1, 2020
Abstract
An inappropriate candidate shortlisted and a potential one missed simply means an inappropriate resume linked to the incorrect keyword. Document classification is being excessively researched upon these days, due to growing interest in text classification which has become a major contributor to the online texts and documents. The repetitive tasks of a person categorizing the details can be handled by the machinery using an expert system that correctly captures and identifies the text and then classifies it into different categories defined. After the preprocessing of the data, the classification is done as a comparative analysis of Bernoulli’s Naïve Bayes, Multinomial Naïve Bayes, Random Forest, Linear SVM and LSVM with elastic penalty classification on the Top 30 Job listing dataset with different parameters and thus we are able to analyze the dependencies between different terms in classes with varying densities and accounts. The accuracy was evaluated and LSVM provides the best accuracy in classifying job entitlements based on the queries submitted and was able to achieve 96.25% accuracy for 55000 samples.
Keywords: TF-IDF, LSVM, BNB, MNB, RF
Suggested Citation: Suggested Citation