Students are increasingly relying on Generative AI (GAI) to support their writing - a key pedagogical practice in education. In GAI-assisted writing, students can delegate core cognitive tasks (e.g., generating ideas and turning them into sentences) to GAI while still producing high-quality essays. This creates new challenges for teachers in assessing and supporting student learning, as they often lack insight into whether students are engaging in meaningful cognitive processes during writing or how much of the essay's quality can be attributed to those processes. This study aimed to help teachers better assess and support student learning in GAI-assisted writing by examining how different writing behaviors, especially those indicative of meaningful learning versus those that are not, impact essay quality. Using a dataset of 1,445 GAI-assisted writing sessions, we applied the cutting-edge method, X-Learner, to quantify the causal impact of three GAI-assisted writing behavioral patterns (i.e., seeking suggestions but not accepting them, seeking suggestions and accepting them as they are, and seeking suggestions and accepting them with modification) on four measures of essay quality (i.e., lexical sophistication, syntactic complexity, text cohesion, and linguistic bias). Our analysis showed that writers who frequently modified GAI-generated text - suggesting active engagement in higher-order cognitive processes - consistently improved the quality of their essays in terms of lexical sophistication, syntactic complexity, and text cohesion. In contrast, those who often accepted GAI-generated text without changes, primarily engaging in lower-order processes, saw a decrease in essay quality. Additionally, while human writers tend to introduce linguistic bias when writing independently, incorporating GAI-generated text - even without modification - can help mitigate this bias.

The recent large language models (LLMs), e.g., ChatGPT, have been able to generate human-like and fluent responses when provided with specific instructions. While admitting the convenience brought by technological advancement, educators also have concerns that students might leverage LLMs to complete their writing assignments and pass them off as their original work. Although many AI content detection studies have been conducted as a result of such concerns, most of these prior studies modeled AI content detection as a classification problem, assuming that a text is either entirely human-written or entirely AI-generated. In this study, we investigated AI content detection in a rarely explored yet realistic setting where the text to be detected is collaboratively written by human and generative LLMs (termed as hybrid text for simplicity). We first formalized the detection task as identifying the transition points between human-written content and AI-generated content from a given hybrid text (boundary detection). We constructed a hybrid essay dataset by partially and randomly removing sentences from the original student-written essays and then instructing ChatGPT to fill in for the incomplete essays. Then we proposed a two-step detection approach where we (1) separated AI-generated content from human-written content during the encoder training process; and (2) calculated the distances between every two adjacent prototypes (a prototype is the mean of a set of consecutive sentences from the hybrid text in the embedding space) and assumed that the boundaries exist between the two adjacent prototypes that have the furthest distance from each other. Through extensive experiments, we observed the following main findings: (1) the proposed approach consistently outperformed the baseline methods across different experiment settings; (2) the encoder training process (i.e., step 1 of the above two-step approach) can significantly boost the performance of the proposed approach; (3) when detecting boundaries for single-boundary hybrid essays, the proposed approach could be enhanced by adopting a relatively large prototype size (i.e., the number of sentences needed to calculate a prototype), leading to a 22% improvement (against the best baseline method) in the In-Domain evaluation and an 18% improvement in the Out-of-Domain evaluation.

Automatic Text Scoring (ATS) is a widely-investigated task in education. Existing approaches often stressed the structure design of an ATS model and neglected the training process of the model. Considering the difficult nature of this task, we argued that the performance of an ATS model could be potentially boosted by carefully selecting data of varying complexities in the training process. Therefore, we aimed to investigate the effectiveness of curriculum learning (CL) in scoring educational text. Specifically, we designed two types of difficulty measurers: (i) pre-defined, calculated by measuring a sample's readability, length, the number of grammatical errors or unique words it contains; and (ii) automatic, calculated based on whether a model in a training epoch can accurately score the samples. These measurers were tested in both the easy-to-hard to hard-to-easy training paradigms. Through extensive evaluations on two widely-used datasets (one for short answer scoring and the other for long essay scoring), we demonstrated that (a) CL indeed could boost the performance of state-of-the-art ATS models, and the maximum improvement could be up to 4.5%, but most improvements were achieved when assessing short and easy answers; (b) the pre-defined measurer calculated based on the number of grammatical errors contained in a text sample tended to outperform the other difficulty measurers across different training paradigms.

The common item-based collaborative filtering framework becomes a typical recommendation method when equipped with a certain item-to-item similarity measurement. On one hand, we realize that a well-designed similarity measurement is the key to providing satisfactory recommendation services. On the other hand, similarity measurements designed for sequential recommendation are rarely studied by the recommender systems community. Hence in this paper, we focus on devising a novel similarity measurement called position-aware similarity (PAS) for sequential recommendation. The proposed PAS is, to our knowledge, the first count-based similarity measurement that concurrently captures the sequential patterns from the historical user behavior data and from the item position information within the input sequences. We conduct extensive empirical studies on four public datasets, in which our proposed PAS-based method exhibits competitive performance even compared to the state-of-the-art sequential recommendation methods, including a very recent similarity-based method and two GNN-based methods.

Zijie Zeng, Xinyu Li, Dragan Gasevic, Guanliang Chen. Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2022.

Exploiting temporal effect has empirically been recognized as a promising way to improve recommendation performance in recent years. In real-world applications, one-class data in the form of (user, item, timestamp) are usually more accessible and abundant than numerical ratings. In this article, we focus on exploiting such one-class data in order to provide personalized next-item recommendation services. Specifically, we base our work on the framework of time-aware item-based collaborative filtering and propose a simple yet effective similarity measurement called bidirectional item similarity (BIS) that is able to capture sequential patterns even from noisy data. Furthermore, we extend BIS via some factorization techniques and obtain an adaptive version, i.e., adaptive BIS (ABIS), in order to better fit the behavioral data. We also design a compound weighting function that leverages the complementarity between two well-known time-aware weighting functions. With the proposed similarity measurements and weighting function, we obtain two novel collaborative filtering methods that are able to achieve significantly better performance than the state-of-the-art methods, showcasing their effectiveness for next-item recommendation.