The 28th International Joint Conference on Artificial Intelligence
August 10-16, 2019
Text generation and automatic writing have gradually been one of the frontiers for the artificial intelligence community. To facilitate the development of the text generation filed, we summarize existing researches and give an overview of their technical implementations in this tutorial. Specifically, we focus on the recent advances and trends in creative and artistic writing, including the genres of story, poetry, couplets.
Outline for the tutorial
- Introduction to artistic writing
- History review
- Recent advances
- Challenges and problems
- Background knowledge
- Concepts, problem formulation, and task statements
- Deep learning for long text generation
- Existing Methods
- Towards better evaluation for artistic writing
- Better training: reinforcement learning and adversarial training
- Recent trends
- Multimodal generation
- Extra knowledge combination
- Controllable generation
- Cross-lingual generation
- Summary and future directions
- Remaining challenges
- Potential solutions
The slides for the tutorial are available here
Automatic writing is one of the common goal and vision for the AI community, which involves various research branches such as narrative intelligence, knowledge representations, text generation, etc. Among different genres of writing, creative and artistic writing (e.g., poetry, story) has attracted enormous interest in both academia and industry due to their values in aesthetic, culture, entertainment, long text generation technology. Specifically, artistic writing generally uses higher quality wording and more specific rules than general writing, which needs to consider both the requirement of long text generation and aesthetic. Thus, this tutorial will attract a considerable group in AI community who are interested in building an entertainment writing system such as poetry generation system, learning the advances of long text generation such as how to address the issue of thematic consistency, seeing what machine can do and cannot do from the perspective of narrative intelligence.
Building upon our long-term efforts on artistic writing, e.g., story generation and poetry composition, this tutorial aims at introducing the “What is the difference between artistic writing and general wiring from the perspective of AI?”, “What are the challenges of these specific genres?”, “How to address these issues in existing models?”, “How to build an artistic writing system?”, “What are the remaining issues and the research trends?”.
Audience and Requirements
Our target audience are researchers and practitioners with some deep learning and text generation background who are interested in artistic writing (e.g., couplet generation, poetry composition, and story writing) and the technologies behind the prosperity of artistic writing in industry and academia, or would like to learn how to build a story/poetry/couplet generation systems with the state-of-the-art technologies.
We have no special requirement for the proposed tutorial.
Juntao Li is now in the 4-th year of his doctoral program at Peking University, supervised by Professor Rui Yan. His research focuses on Natural Language Processing and Artificial Intelligence. More concretely, he is now working on personalized conversation systems, data augmentation for text generation, and artistic writing (i.e., poetry generation, story generation, etc.).
Dr. Rui Yan is now a tenure-track assistant professor at Peking University, and he is affiliated with Data Science Center, Beijing Institute of Big Data Research. He is also an adjunct professor at Central China Normal University and Central University of Finance and Economics. He was a Senior Researcher at Baidu Inc. For the past 10+ years, Dr. Rui Yan has been working on Artificial Intelligence (AI) for Natural Language Processing (NLP) and other related research fields such as Data Mining (DM), Information Retrieval (IR) and Machine Learning (ML). Dr. Rui Yan now focus on human-computer conversational models (a.k.a., dialogues), natural language cognition and generation, summarization and other interdisciplinary tasks. Previously, he has been invited to give tutorial talks at EMNLP, WWW and SIGIR.