Large Language Models (LLMs) have achieved state-of-the-art performance in Aspect Sentiment Quadruplet Prediction (ASQP). However, we argue that this success often relies on superficial positional heuristics rather than robust structural reasoning. In this paper, we propose a Probe-and-Cure framework to scrutinize and enhance LLM robustness. First, we introduce a Multi-Surgical Adversarial Attack protocol as a diagnostic tool. Our study reveals that SOTA models are “Fragile Giants”: they suffer severe performance degradation when facing logical distractors and deep syntactic embeddings. To address this, we propose Adaptive Bi-SADA (Bidirectional Structure-Aware Data Augmentation). Unlike uniform augmentation strategies, our method constructs a length-aware curriculum: it applies natural structural hardening to short sentences to prevent overfitting, and syntactic normalization to long sentences to distill core dependencies. We implement a strict generation-time verification protocol to ensure semantic invariance. Experiments on ASQP and ACOS tasks demonstrate that our method not only achieves new SOTA F1 scores but also effectively transforms superficial pattern matching into robust structural reasoning, significantly closing the performance gap under adversarial stress tests.

Cognitive Diagnosis Models (CDMs) play a pivotal role in personalized online learning. Traditional CDMs rely on discrete, ID-based embeddings to represent students, exercises, and concepts. This paradigm diverges from the nature of learner cognition, where knowledge is not stored and retrieved as isolated symbols. As a result, CDMs suffer from semantic limitations when new exercises or concepts appear. In this paper, we propose a Process-aware Language Cognitive Diagnosis (PLCD) framework that uses language-derived structures as cognitive priors and response records to calibrate student posterior states. PLCD leverages large language models (LLMs) to construct concept schemas and cognitive process graphs, and uses target-conditioned semantic memory to retrieve historical responses that are relevant to each target exercise. A process-grounded Language-to-Cognition Mapper with DA-MoE experts and process-level contrastive learning then maps the textual evidence into a unified cognitive space. Experimental results show that PLCD not only outperforms traditional baselines in predicting student performance but also exhibits strong cognitive transfer capabilities. These results connect the computational power of LLMs with the psychometric goal of measuring latent knowledge states, suggesting that structured language priors calibrated by response records can improve cold-start robustness and cognitive grounding.

Bilingual Lexicon Induction (BLI) is the task of translating words from corpora in two languages. Recent advances in BLI work by aligning the two word embedding spaces. Following that, a key step is to retrieve the nearest neighbor (NN) in the target space given the source word. However, a phenomenon called hubness often degrades the accuracy of NN. Hubness appears as some data points, called hubs, being extra-ordinarily close to many of the other data points. Reducing hubness is necessary for retrieval tasks. One successful example is Inverted SoFtmax (ISF), recently proposed to improve NN. This work proposes a new method, Hubless Nearest Neighbor (HNN), to mitigate hubness. HNN differs from NN by imposing an additional equal preference assumption. Moreover, the HNN formulation explains why ISF works as well as it does. Empirical results demonstrate that HNN outperforms NN, ISF and other state-of-the-art. For reproducibility and follow-ups, we have published all code.