Large language models (LLMs) have become an important semantic infrastructure for modern recommender systems. A prevailing paradigm integrates LLM-derived semantic embeddings with collaborative representations via representation alignment, implicitly assuming that the two views encode a shared latent entity and that stronger alignment yields better results. We formalize this assumption as the global low-complexity alignment hypothesis and argue that it is stronger than necessary and often structurally mismatched with real-world recommendation settings. We propose a complementary perspective in which semantic and collaborative representations are treated as partially shared yet fundamentally heterogeneous views, each containing both shared and view-specific factors. Under this shared-plus-private latent structure, enforcing global geometric alignment may distort local structure, suppress view-specific signals, and reduce informational diversity. To support this perspective, we develop complementarity-aware diagnostics that quantify overlap, unique-hit contribution, and theoretical fusion upper bounds. Empirical analyses on sparse recommendation benchmarks reveal low item-level agreement between semantic and collaborative views and substantial oracle fusion gains, indicating strong complementarity. Furthermore, controlled alignment probes show that low-capacity mappings capture only shared components and fail to recover full collaborative geometry, especially under distribution shift. These findings suggest that alignment should not be treated as the default integration principle. We advocate a shift from alignment-centric modeling to complementarity fusion-centric, complementarity-aware design, where shared factors are selectively integrated while private signals are preserved. This reframing provides a principled foundation for the next generation of LLM-enhanced recommender systems.

With the explosive adoption of web-based technology, the amount of graph-structured data has increased dramatically, resulting in a higher demand to find anomalous patterns in different types of online services, such as fraudulent transactions, fake accounts, and coordinated malicious campaigns. The performance of anomaly detection in graphs of web systems is a challenge due to the sparse and camouflaged nature of such anomalies, multi-granular irregularity features, and the instability of the generative models in a real-world web application. To address these constraints, we introduce a new unified generative framework GRAND (Graph Anomaly Detection via Diffusion) suitable for graph data of the web domain. GRAND applies a novel dual-diffusion strategy: continuous Gaussian diffusion for node features and discrete diffusion for edges, combined with a structural-prior-conditioned graph transformer denoiser. Besides this, the framework adds strong anomaly scoring mechanisms with adaptive pooling and normalization schemes to detect the subtle anomaly signals typical of web data. It provides degeneracy detection for inference stability. GRAND has been demonstrated to obtain better results than state-of-the-art methods by extensive evaluations of different benchmark datasets. GRAND demonstrates strong performance in the detection of money laundering, particularly on the Elliptic dataset, a Bitcoin transaction graph typically for financial web applications. © 2026 Owner/Author.