Construction of a Task Ontology for Informatization of Job Search Activities
Kodai Matsuda (AY 2021)
Every year, many university students engage in job search activities. The number of job seekers in 2019 is 446,882. The average per capita cost of hiring new graduates in the year 2018 is 726,000. Thus, since job-hunting activities have social and economic impacts, the results of research on job-hunting activities are considered to have social and economic impacts. Research on the phenomenon of job search activities requires resources to understand job search activities objectively, but there have been no scientific attempts to systematize the phenomenon of job search activities into an informative format. Informatization here refers to representing and formalizing job search phenomena so that they can be treated as concepts.
In this study, we collected data on entry sheet (hereinafter referred to as ES) preparation procedures from books on job search and interviews with people who have experienced job search, and based on these data, we constructed a task ontology to inform the phenomenon of job search. A task ontology is a set of concepts and relationships among concepts to describe the problem-solving activity of writing an ES that university students perform in their job search.
When collecting data from books, we extracted object-verb pairs from the text of the books. A total of 109 data were extracted from 16 pages describing specific job search phenomena. In the interview data collection, ten subjects, consisting of eight university seniors and two graduate school sophomores who had received job offers, were asked about the procedure of creating an ES in a one-hour online manual creation experiment. The resulting data consisted of a total of 193 objects and verbs. The collected data were then used to construct a task ontology. The Activity-First Method (hereafter AFM) was used as the method for constructing the task ontology, since AFM is designed to construct ontologies from manuals and thus has an affinity with the collected data.
Through the construction of the task ontology in this study, we have shown that the phenomenon of job search can be informed by basic concepts and role concepts using a task ontology. Role concepts are concepts that depend on context and other concepts. The task ontology includes the actions of job seekers, terms used in job search, and objects that exist in the real world. For example, what is expressed as "showcasing one's strengths" in a book can be transformed into "writing an event with the role of showing one's strengths on an artifact in the form of text" by using the task ontology. By presenting the basic concepts of events and letters, it is expected that the task will be more comprehensible to new job seekers who do not know the context.
Future directions include the creation of a system that automatically collects the job search tasks that were manually collected in this study, the development of an agent system that supports job seekers in their job search activities through interactive communication, and the study of a job search support method that generates tasks from the task ontology and presents them to job seekers. The study of a job hunting support method that generates tasks from a task ontology and presents them to job hunters is also considered.
(Translated by DeepL)