Search

Recently, more and more people study online for the convenience of access to massive learning materials (e.g. test questions/notes), thus accurately understanding learning materials became a crucial issue, which is essential for many educational applications. Previous studies focus on using language models to represent the question data. However, test questions (TQ) are usually heterogeneous and multi-modal, e.g., some of them may only contain text, while others half contain images with information beyond their literal description. In this context, both supervised and unsupervised methods are difficult to learn a fused representation of questions. Meanwhile, this problem cannot be solved by conventional methods such as image caption, as the images may contain information complementary rather than duplicate to the text. In this paper, we first improve previous text-only representation with a two-stage unsupervised instance level contrastive based pre-training method (MCL: Mixture Unsupervised Contrastive Learning). Then, TQ-Net was proposed to fuse the content of images to the representation of heterogeneous data. Finally, supervised contrastive learning was conducted on relevance prediction-related downstream tasks, which helped the model to learn the representation of questions effectively. We conducted extensive experiments on question-based tasks on large-scale, real-world datasets, which demonstrated the effectiveness of TQ-Net and improve the precision of downstream applications (e.g. similar questions +2.02% and knowledge point prediction +7.20%). Our code will be available, and we will open-source a subset of our data to promote the development of relative studies., Comment: This paper has been accepted for the AAAI2023 AI4Edu Workshop

Cross domain sentiment classification (CDSC) aims to adopt a model trained by a source domain to a target domain. It has received considerable attention in recent years. Most existing models mainly focus on learning representations that are domain independent in both the source domain and the target domain. However, domain specific features, which should also be informative are ignored by these models. In this paper, we propose an end to end model. It can capture both the source domain and target domain features at the same time. This model includes two parts; one is a cloze task network (CTN), we use it as an auxiliary task to fine-tune words embedding in both domains. Another is a Convolutional hierarchical attention networks (CHAN), we use it for sentiment classification. The CHAN can capture important words and sentences concerning sentiment based on its two stages of attention mechanism. The CTN and CHAN conduct jointly learning we abbreviate this model as CCHAN. The experiments on the Amazon review datasets demonstrate that the proposed CCHAN can significantly outperform the state-of-the-art methods.