Article Preview
TopIntroduction
As the backbone of the country’s and society’s future development, college students will face severe consequences if psychological factors affect their development (Vandebroek et al., 2018). In terms of their interpersonal environment, entering high school from junior high school, students face new teachers and classmates and a new learning environment and school management atmosphere; boarding students also face new living environments and conditions, as well as new psychological weaning (Yan et al., 2022). University teachers and students increasingly recognize song therapy. This study uses group song therapy, drawing on relevant theories of group psychological counseling, to systematically intervene in the psychological stress of college students and explore the application effect of group song therapy on stress reduction in college students. However, existing song fragment melody retrieval methods continue to be based on the physical characteristics of the song, and the retrieval process cannot continue when users do not remember the song melody (Wang, 2014). In the advanced semantic features of songs, emotional semantics are higher-level features beyond melodic semantics. Therefore, users’ emotional needs must be considered thoroughly in song retrieval. At the same time, the users of emotional songs are not experts but ordinary users, which results in songs—searched based on input emotional characteristics—not being what users need (Li & Yu, 2021).
On this basis, songs must be labeled with the emotions of ordinary users to make comprehensive recommendations in combination with expert categories. On professional forums, the classification of songs is accurate and detailed, providing users with a friendly interactive platform, and the emotional tendencies reflected by a large number of user comments determine the public emotions of a song. Therefore, this article proposes a song sentiment semantic analysis, processing, and retrieval scheme based on song comment content. The article uses a clustering algorithm-based outlier analysis algorithm to analyze these data in detail, attempting to determine the mental health status of each college student. Then, we use the Apriori association rule algorithm to explore the intrinsic connections between psychological problems and various attributes, thereby providing a reliable basis for decision-makers. The song’s emotions are taken from the user and fed back to the user, thus achieving the effect that the song’s emotions are closely related to the user and change with the user. Through this study, we hope to provide more effective psychological intervention methods for college students to promote healthy growth and development.
This paper mainly offers the following innovations:
- 1.
If mental illness is regarded as “abnormal,” it can be regarded as a “rare category” or “outlier” due to the small number of such categories, which can be classified as the problem of mining unbalanced data sets. For this kind of problem, it is difficult to achieve ideal results when using traditional clustering and classification techniques (categories must be known in advance).
- 2.
Most research on college students’ psychological problems collects data through mental health testing tools (such as the SCL-90 symptom self-rating scale or the Chinese College Students’ Mental Health Assessment System). Then, it conducts analysis and statistics on these data. The content of the psychological test questionnaire is effective for the judgment and prediction of mental health. The conclusion is scientific and accurate, but this method is still lacking in determining the cause analysis of mental health problems. It does not consider the influence of many factors on mental health, such as the fluctuation of students’ academic performance, family status, and students’ love. Therefore, it is more reasonable and effective to comprehensively use various data and apply correlation analysis techniques to mine and analyze the data.