Developing Alumni Job Prediction Models Based on Term Occurrences and Word Class Analysis of Search Engine Page Results

Toba, Hapnes and Wijaya, Evelyn A. and Wijanto, Maresha Caroline and Karnalim, Oscar (2016) Developing Alumni Job Prediction Models Based on Term Occurrences and Word Class Analysis of Search Engine Page Results. World Transactions on Engineering and Technology Education, 14 (4). pp. 563-567. ISSN 1446-2257

[img]
Preview
Text
3.pdf - Published Version

Download (447Kb) | Preview

Abstract

In this study, the authors propose several prediction models for alumni, which are based on term occurrences, word class analysis, such as nouns and verbs, and the clusters aggregation method. During the data collection process, a clustering-based method to disambiguate alumni names from the search engine page results was implemented. Further, several job prediction classifiers based on word occurrences, verbs-nouns word classes were developed and a cluster-based aggregation method. The classifiers were evaluated by using a real alumni tracer study, and tested in a cross-validation (5-fold) and hold-out combinations. The experimental results revealed that in the engineering and economics fields, there are more variations of alumni professions. Based on the experimental data, the approach seems to be useful for finding people in general, especially, for ordinary alumni names. As future work, it would be interesting to explore various social media, beyond LinkedIn. Moreover, friendship relations in social media, such as Facebook and Twitter, might also be beneficial for further alumni relational analysis.

Item Type: Article
Subjects: T Technology > T Technology (General)
Depositing User: Perpustakaan Maranatha
Date Deposited: 20 Dec 2017 06:51
Last Modified: 10 Oct 2018 08:46
URI: http://repository.maranatha.edu/id/eprint/23794

Actions (login required)

View Item View Item