Answering questions based on a reference text is a frequently employed comprehension assessment method that enables teachers to effectively and efficiently evaluate students. Various tools and methods were developed to tackle automated question generation, however, selecting valid answer candidates as a first step is less addressed. Thus, we introduce a solution built on top of FairytaleQA and tailored for training a DeBERTa-based model to classify the quality of each candidate to be part of a strong answer-question pair. First, we extract answer candidates by syntactically parsing the context (i.e., selecting text spans from the reference text based on the nodes in the constituency tree); then, questions are generated for the extracted candidates using a pre-trained LLM model on this task. Next, we assess a candidate’s quality by relying on another fine-tuned model’s capability to answer the previously generated question for that candidate. This enables us to categorize answers using a four-class system: very good, good, average, and unusable. A significant advantage of our method is that the encoder classifier can score all potential answer candidates in a single inference step for the entire context. We compare our selection against both the answers from explicit questions in the original dataset and a fine-tuned LLM for answer selection using an Elo ranking system. In addition, we propose three strategies based on semantic similarity and text position to ensure coverage and diversity of candidates’ selection.

Question generation plays an important role in educational applications, enabling automated assessment and reading comprehension support. Attribute-controlled question generation aims to produce questions that fit predefined characteristics such as difficulty, focus, or coverage. Existing methods predominantly rely on supervised fine-tuning, which often fails to impose a strong adherence to attribute values, resulting in weak coupling between prompt specifications and model outputs. We introduce Odds-Ratio Steerable Optimization (ORSO), a framework designed to enhance attribute sensitivity in question generation models. Building upon preference-based learning techniques without requiring human-curated preference sets, ORSO employs input-level perturbations to create contrastive training signals. Empirical evaluations on both exhaustive and expert-validated attribute configurations indicate that ORSO performs better in enforcing attribute conformity while maintaining output quality. These results argue for the benefits of explicit attribute-aware optimization in controllable question generation tasks.

Writing development is often assessed through aggregate improvements in surface-level features, yet less attention has been given to how multiple linguistic dimensions evolve jointly over time. We model writing development as a multidimensional system shaped by stable individual variation and instructional progression across staged assignments, using interpretable linguistic features from the Writing Analytics Toolkit (WAT) and transformer-based sentence embeddings.Variance partitioning reveals substantial between-student stability alongside stage-dependent change. Mixed-effects models identify non-uniform developmental trajectories: academic focus, information density, and conventional language increase, whereas development of ideas and lexical variety decline, indicating tradeoffs across competing dimensions. Cross-lagged analyses further show dynamic dependencies between dimensions, suggesting coordinated change rather than independent progression.Embedding-based analyses capture stage-dependent shifts in semantic representation, with larger changes in earlier stages and increasing stability over time. Although assignment structure contributes to observed variation, stable individual differences and cross-stage dependencies indicate underlying developmental processes that generalize across tasks.Together, these findings characterize writing development as structured change in a multidimensional representational system, highlighting the need for computational models that capture stable variation, non-monotonic trajectories, and interactions among linguistic components.

Research in AI applied to education increasingly relies on large-scale, high-quality datasets to support the development and evaluation of learning analytics and intelligent educational systems. Open educational resources provide a promising foundation, yet few datasets integrate structured instructional content with assessment materials in a multimodal form. In this study, we introduce a large-scale multimodal educational dataset (EduMUSE - Educational Multimodal Understanding & Solution Dataset) constructed from OpenStax undergraduate textbooks across multiple domains. The dataset integrates hierarchically structured instructional text, figures, exercises, and, when available, official solutions. For exercises with solutions, we introduce an automatic method that associates each exercise with a focused instructional subsection rather than entire textbook chapters, estimating subsection relevance via solution likelihood under candidate contexts using a vision–language model. We analyze the impact of contextualization on the behavior of vision–language models across different contexts. Results indicate that subsection-level instructional context has a measurable impact on model performance, with variation across model scales and task formulations. The dataset and code are released as open source at https://github.com/upb-nlp/BEA-EduMUSE/ to support reproducible research in multimodal educational modeling and to facilitate generating similar datasets using our approach.