Method for Clustering Comments in Class Evaluation Questionnaires using Keyword Feature Scores

In Japanese universities, improving classes is a task that must be continually implemented. Broadly speaking, a university class’s purpose is threefold: (1) have students understood the content of the class, (2) have they achieved their goals for the class, and (3) were they satisfied with the class...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:International journal of education (Las Vegas, Nev.) Nev.), 2024-12, Vol.16 (4), p.1
Hauptverfasser: Iwano, Maya, Tsuda, Kazuhiko
Format: Artikel
Sprache:eng
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:In Japanese universities, improving classes is a task that must be continually implemented. Broadly speaking, a university class’s purpose is threefold: (1) have students understood the content of the class, (2) have they achieved their goals for the class, and (3) were they satisfied with the class? Universities must embrace these three points. For this reason, class evaluation questionnaires administered at many universities nearly always include questions on “comprehension,” “achievement,” and “satisfaction.” However, it is not possible to collect sufficient information by simply establishing questions and rating the responses. Therefore, research analyzing free descriptions in class evaluation questionnaires is increasing. This study developed a system to classify free descriptions provided by students into “comprehension,” “achievement,” and “satisfaction.” In addition, it proposed a method of knowledge construction for such classification. One of its benefits was that it could be implemented within a short period of time with little effort. The proposed method’s accuracy was evaluated by comparing the results of the classification by means of the proposed method with the results of the classification of the class evaluation questionnaire by an analyst. The two classification results matched with a probability of 92.67% to 93.67%, confirming that the method was sufficiently practical.
ISSN:1948-5476
1948-5476
DOI:10.5296/ije.v16i4.22097