Hybrid occupation recommendation for adolescents on interest, profile, and behavior
Introduction
Students make initial but critical decisions regarding what to study and which career path to pursue (Ferry, 2006). A mass of students has decided their majors/occupations out of proper and professional advice from school services. Mismatch of the major choice and lack of information through professional study is one of the reasons for them to change major. Such changes are wasteful in time and resources, and it is the cause of financial and emotional stresses of students. An approximated twenty to fifty percent of students enter college as undecided major; and an approximated seventy-five percent of students change their major at least once before graduation (Gordon, 2007). On the other hand, the students’ major choices are influenced by the society, the education environment, and mostly their families. Those pitfalls are potentially the causes of a mismatch major between academic achievements, personality, interest and abilities of students. It would be useful to understand how students’ choice of the academic majors depends on personal characteristics, competencies, and vocational interests. Most of the students do not possess adequate information about meaning of occupations/majors, what careers can be reached by which majors, and what kind of skills and abilities are needed for a particular occupation/major. Especially, even some parents are unfamiliar with the contemporary occupations/majors.
The main goal of this paper is to construct an occupation recommendation system by using data mining method. In the experiment, we focus on testing Mongolian, Sri Lankan, Taiwanese and Thai students. The system can provide details of occupations and can assist the students for major choices, as well as the careers to pursue. Furthermore, the research goal incorporates a set of results, which are recommended using similarity measurements and recommendation techniques. We called this a hybridization system. These methods serve as a base for recommending occupations that meet interests and competencies of students.
The rest of the paper is organized as follows. Section 2 presents related works about Content-based, Collaborative filtering, and hybrid techniques. In Section 3, we introduce the proposed system its methods and functions. Section 4 provides experimental results. Finally, Section 5 contains the conclusion and future work.
Section snippets
Related work
Recommendation systems are tools that use user information (or another user who have the same behavior) and prioritize items likely to be of interest to a user. Recommender systems are applied in a variety of application such as movies, music, books, and products. Many websites that we use every day are designed based on some recommendation algorithms (Burke et al., 2011). Amazon, the popular e-commerce site, uses content-based recommendation. When a user chooses an item to buy, Amazon
The proposed system
OCCupation RECommendation (OCCREC) is a website (http://occrec.minelab.tw/) that helps to discover the occupation interest. The students who want to know about occupation information can search and also share the result based on the questionnaires on Facebook. A general illustration of methods is shown in Fig. 1. OCCREC consist of three main part:
First, OCCREC collects personal details from the user by filling the form or logging-in with Facebook and then the system display a list of
Results and discussion
In this section, we present our experimental results of applying hybrid techniques for recommending the occupations to the students. Data collection is from our research website at http://occrec.minelab.tw.
Conclusion
In this paper, we implement the experiments with 612 students from Mongolia, Sri Lanka, Taiwan, and Thailand in different study fields. In order to recommend the well-fitting occupation to students using their vocational interest, we employed the hybrid recommendation algorithms. We compared five kinds of similarities, namely, Euclidean, Intersection, Jaccard, Pearson, and Cosine. Pearson distance has lower MAE and RMSE has the better result (user satisfaction on feedback) in MRR. This finding
References (36)
- et al.
Studies of the hexagonal model: an evaluation: or the perils of stalking the perfect hexagon. Special Issue: Holland’s theory
J. Vocational Behav.
(1992) - et al.
A survey of job recommender systems
Int. J. Phys. Sci.
(2012) - et al.(1999)
- Bolormaa, B., Oyunsuren, B., Altangerel, C., Tsolmon, C., 2016. Akhlakh angiin suragchdiin mergejil songoltin...
- et al.
LinkedVis: exploring social and semantic career recommendations
Hybrid recommender systems: survey and experiments
User Model. User-adapted Interact.
(2002)- et al.
Recommender systems: an overview
AI Mag.
(2011) - et al.
Toward the next generation of recruitment tools: an online social network-based job recommender system
- et al.
Taxonomy-based job recommender systems on Facebook and LinkedIn profiles
- Enkhtuvshin, S., 2013. Suraltsagchdad ajil mergejliin chig barimjaa olgoj bui baidaliin sudalgaa. [The research on...