Zhenjie Zhao

Zhenjie Zhao, Yufang Hou, Dakuo Wang et al.

Generating educational questions of fairytales or storybooks is vital for improving children's literacy ability. However, it is challenging to generate questions that capture the interesting aspects of a fairytale story with educational meaningfulness. In this paper, we propose a novel question generation method that first learns the question type distribution of an input story paragraph, and then summarizes salient events which can be used to generate high-cognitive-demand questions. To train the event-centric summarizer, we finetune a pre-trained transformer-based sequence-to-sequence model using silver samples composed by educational question-answer pairs. On a newly proposed educational question answering dataset FairytaleQA, we show good performance of our method on both automatic and human evaluation metrics. Our work indicates the necessity of decomposing question type distribution learning and event-centric summary generation for educational question generation.

Zhenjie Zhao

Live animation of 2D characters has recently become a popular way for storytelling, and has potential application scenarios like tele-present agents or robots. As an extension of human-human communication, there is a need for augmenting the emotional communication experience of live animation. In this paper, we explore the emotional expressiveness issue of 2D live animation. In particular, we propose a descriptive emotion command model to bind a triggering action, the semantic meaning, psychology measurements, and behaviors of an emotional expression. Based on the model, we designed and implemented a proof-of-concept 2D live animation system, where a novel visual programming tool for editing the behaviors of 2D digital characters, and an emotion command recommendation algorithm are proposed. Through a user evaluation, we showcase the usability of our system and its potential for boosting creativity and enhancing the emotional communication experience.

Zhenjie Zhao

Virtual museums aim to promote access to cultural artifacts. However, they often face the challenge of getting audiences to read and understand a large amount of information in an uncontrolled online environment. Inspired by successful practices in physical museums, we investigated the possible use of guiding questions to engage audiences in virtual museums. To this end, we first identified how to construct questions that are likely to attract audiences through domain expert interviews and mining cultural-related posts in a popular question and answer community. Then in terms of the proactive level for attracting users' attention, we designed two mechanisms to interactively prompt questions: active and passive. Through an online experiment with 150 participants, we showed that having interactive guiding questions encourages browsing and improves content comprehension. We discuss reasons why they are useful by conducting another qualitative comparison and obtained insights about the influence of question category and interaction mechanism.