Name
Abstract | ||
|---|---|---|
Nowadays, college students are prone to feel lonely, thus loneliness identification is an essential task. Existing methods like questionnaire are mainly used to identify loners through loneliness scales. These methods are subjective and often ineffective because loners may avoid reporting their real conditions. In this paper, we propose a new method based on pair-wise course collaborative relationships for identifying lonely students in an end-to-end fashion. Due to the overlook of relations among instances, traditional machine learning methods are insufficient to distill collaborative information from course records. In order to make full use of the interaction among students in course, we use weighted Graph Neural Networks (GNNs) to model our problem. The number of times that two students have collaborated is extracted as edge weight to build a weighted graph. Meanwhile, we propose a Binary tree-based Graph Oversampling Algorithm (BGOA) to tackle class-imbalanced problem. The experimental results have proved that the proposed approach can achieve above 88% of performance. |
Year | DOI | Venue |
|---|---|---|
2020 | 10.1109/ICTAI50040.2020.00080 | 2020 IEEE 32nd International Conference on Tools with Artificial Intelligence (ICTAI) |
Keywords | DocType | ISSN |
loneliness identification,non-Euclidean data,graph neural networks,machine learning,social networks,oversampling,class-imbalanced problem | Conference | 1082-3409 |
ISBN | Citations | PageRank |
978-1-7281-8536-1 | 0 | 0.34 |
References | Authors | |
9 | 4 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Qing Zhou | 1 | 87 | 11.91 |
| Jiang Li | 2 | 0 | 1.69 |
| Yinchun Tang | 3 | 0 | 0.34 |
| Huan Wang | 4 | 0 | 0.34 |