Abstract:Quality-of-Service (QoS) prediction is a critical task in the service lifecycle, enabling precise and adaptive service recommendations by anticipating performance variations over time in response to evolving network uncertainties and user preferences. However, contemporary QoS prediction methods frequently encounter data sparsity and cold-start issues, which hinder accurate QoS predictions and limit the ability to capture diverse user preferences. Additionally, these methods often assume QoS data reliability, neglecting potential credibility issues such as outliers and the presence of greysheep users and services with atypical invocation patterns. Furthermore, traditional approaches fail to leverage diverse features, including domain-specific knowledge and complex higher-order patterns, essential for accurate QoS predictions. In this paper, we introduce a real-time, trust-aware framework for temporal QoS prediction to address the aforementioned challenges, featuring an end-to-end deep architecture called the Hypergraph Convoluted Transformer Network (HCTN). HCTN combines a hypergraph structure with graph convolution over hyper-edges to effectively address high-sparsity issues by capturing complex, high-order correlations. Complementing this, the transformer network utilizes multi-head attention along with parallel 1D convolutional layers and fully connected dense blocks to capture both fine-grained and coarse-grained dynamic patterns. Additionally, our approach includes a sparsity-resilient solution for detecting greysheep users and services, incorporating their unique characteristics to improve prediction accuracy. Trained with a robust loss function resistant to outliers, HCTN demonstrated state-of-the-art performance on the large-scale WSDREAM-2 datasets for response time and throughput.
What problem does this paper attempt to address?
The main problems that this paper attempts to solve are the problems of data sparsity, cold - start problems, data credibility (such as the existence of outliers and grey - sheep users/services), and temporal dynamic characteristics encountered in Quality of Service (QoS) prediction. Specifically, these problems include:
1. **Insufficient data**:
- **Limited user - service interactions**: Due to the limited interactions between users and services, the QoS data is very sparse. How can these sparse data be effectively utilized for reliable QoS prediction?
- **Cold - start problem**: Newly - added users or services have almost no interaction data. How can the QoS of these new users or services be effectively predicted in the absence of historical data?
2. **Data credibility**:
- **Outlier problem**: Dynamic factors (such as limited network bandwidth or competition among service providers) may lead to outliers in QoS data. How can these outliers be effectively suppressed to ensure reliable QoS prediction?
- **Grey - sheep problem**: Some users or services may exhibit unique QoS invocation patterns, which deviate significantly from other users or services, known as grey - sheep. How can these grey - sheep characteristics be detected and utilized to improve prediction accuracy?
3. **Data representation problems**:
- **Domain - specific features**: How can domain - specific features (such as user preferences and service characteristics) be effectively used to improve the stability, speed, and accuracy of prediction?
- **High - order features**: In addition to domain - knowledge features, exploring high - order features such as complex spatial and temporal granularities is crucial for better data representation learning. How can these complex features be utilized to enhance QoS prediction?
To solve the above problems, the author proposes a real - time, anomaly - resistant temporal QoS prediction framework based on the Hypergraph Convoluted Transformer Network (HCTN). HCTN effectively copes with the challenges of high sparsity, cold - start, grey - sheep problems, and temporal dynamic characteristics by combining the hypergraph structure and graph convolution to handle high - order correlations and using the Transformer network to capture fine - grained and coarse - grained dynamic patterns.
### Formula summary
- **Non - negative Matrix Decomposition (NMD)**:
\[
Q_t^{n\times m}(i, j)\approx(X_u^t)^{n\times f_1}\cdot(X_s^t)^{m\times f_1}_T
\]
where \(Q_t\) is the QoS invocation matrix at the \(t\) - th time step, and \(X_u^t\) and \(X_s^t\) are the low - rank latent feature matrices of users and services respectively.
- **Normalized adjacency matrix**:
\[
\hat{A}_t=(D_t)^{- 1/2}\cdot(A_t + I)\cdot(D_t)^{-1/2}
\]
where \(D_t\) is a diagonal matrix calculated from \(A_t\), and \(I\) is an identity matrix.
- **Hypergraph Convolutional Network (HCN) output aggregation**:
\[
fw(\bar{X})=X_0+\sum_{i = 1}^{l}\frac{1}{i + 1}X_i
\]
where \(\bar{X}=(X_0,X_1,\ldots,X_l)\) is a convolution output tuple of length \(l + 1\).
- **Grey - sheep Difference Index (GDI)**:
\[
GDI(u_i)=\frac{\sum_{j = 1}^{m}|Q_t(i, j)-\mu_t(u_i)|}{\sum_{j = 1}^{m}\mu_t(u_i)}
\]
\[
GDI(s_j)=\frac{\sum_{i = 1}^{n}|Q_t(i, j)-\mu_t(s_j)|}{\sum_{i = 1}^{n}\mu_t(s_j)}
\]
- **Grey - sheep label**:
\[
G(i,1,t)=