期刊名称：International Journal of Computer Science & Technology
印刷版ISSN：2229-4333
电子版ISSN：0976-8491
出版年度：2016
卷号：7
期号：4
页码：111-118
语种：English
出版社：Ayushmaan Technologies
摘要：The main focus is determine a machine learning model for mapping graduates’ skills to industry roles using skills profile of employed graduates. A hierarchical classification strategy using a bottom-up approach was designed based on a taxonomy that is bottom-up friendly and was applied to construct the model. Two machine learning techniques, naiveBayes and support vector machines, and software engineering employees’ profile dataset with 113 instances and 18 attributes were adopted in the investigation using experimental design. Experiments to evaluate the model were designed using pretest-posttest with control group. While the aim was to assess performance of the model under effect of various machine learning techniques and taxonomic structures, performance reported on carefully selected benchmark on bottomup multi-classification method was adopted for validation. Findings indicate model performance is not only considerably fair both under naïve Bayes (57.85%) and SVM (67.15%) but also slightly above the reported benchmark score of 61%. However, difference between the two model designs is significant (t=2.602, p=.029; t= -2.939, p=.017). In conclusion, automatic mapping of graduates’ skills to industry roles with the aim to improve employability and productivity prediction of new graduates must involve both a suitable machine learning technique and a bottomup friendly taxonomic structure.
关键词：Automatic Mapping of Skills;Long Term Unemployment; Machine Learning;Software Engineering
