Challenges and Opportunities in Text Generation Explainability

被引:0
|
作者
Amara, Kenza [1 ]
Sevastjanow, Rita [1 ]
El-Assady, Mennatallah [1 ]
机构
[1] Swiss Fed Inst Technol, Dept Comp Sci, Zurich, Switzerland
关键词
Explainability; Text generation; Autoregressive models; Evaluation; Perturbation-based analysis; Challenges and opportunities;
D O I
10.1007/978-3-031-63787-2_13
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
The necessity for interpretability in natural language processing (NLP) has risen alongside the growing prominence of large language models. Among the myriad tasks within NLP, text generation stands out as a primary objective of autoregressive models. The NLP community has begun to take a keen interest in gaining a deeper understanding of text generation, leading to the development of model-agnostic explainable artificial intelligence (xAI) methods tailored to this task. The design and evaluation of explainability methods are non-trivial since they depend on many factors involved in the text generation process, e.g., the autoregressive model and its stochastic nature. This paper outlines 17 challenges categorized into three groups that arise during the development and assessment of attribution-based explainability methods. These challenges encompass issues concerning tokenization, defining explanation similarity, determining token importance and prediction change metrics, the level of human intervention required, and the creation of suitable test datasets. The paper illustrates how these challenges can be intertwined, showcasing new opportunities for the community. These include developing probabilistic word-level explainability methods and engaging humans in the explainability pipeline, from the data design to the final evaluation, to draw robust conclusions on xAI methods.
引用
收藏
页码:244 / 264
页数:21
相关论文
共 50 条
  • [21] Challenges and opportunities in the third-generation biorefinery
    Shi, Shuobo
    Wang, Yubo
    Qiao, Weibo
    Wu, Longhao
    Liu, Zihe
    Tan, Tianwei
    CHINESE SCIENCE BULLETIN-CHINESE, 2023, 68 (19): : 2489 - 2503
  • [22] Next generation optical fibers: Challenges and opportunities
    Mazzali, Claudio
    ICTON 2007: Proceedings of the 9th International Conference on Transparent Optical Networks, Vol 1, 2007, : 313 - 316
  • [23] Challenges and opportunities in next generation of electropolishing surfaces
    Vara, Gemma A.
    Jesus Butron, E.
    BelenGarcia-Blanco, M.
    SURFACE ENGINEERING, 2015, 31 (06) : 397 - 398
  • [24] Next generation battery technology, challenges, and opportunities
    Visco, Steven J.
    Nimon, Eugene
    De Jonghe, Lutgard
    Katz, Bruce
    Chu, May-Ying
    ABSTRACTS OF PAPERS OF THE AMERICAN CHEMICAL SOCIETY, 2013, 245
  • [25] Arabic text detection: a survey of recent progress challenges and opportunities
    Muaad, Abdullah Y.
    Raza, Shaina
    Naseem, Usman
    Davanagere, Hanumanthappa J. Jayappa
    APPLIED INTELLIGENCE, 2023, 53 (24) : 29845 - 29862
  • [26] EMOTIONALLY WRAPPED SOCIAL MEDIA TEXT: APPROACHES, OPPORTUNITIES, AND CHALLENGES
    Rawat, Tara
    Jain, Shikha
    SCALABLE COMPUTING-PRACTICE AND EXPERIENCE, 2023, 24 (04): : 797 - 818
  • [27] Text-based emotion detection: Advances, challenges, and opportunities
    Acheampong, Francisca Adoma
    Chen Wenyu
    Nunoo-Mensah, Henry
    ENGINEERING REPORTS, 2020, 2 (07)
  • [28] Arabic text detection: a survey of recent progress challenges and opportunities
    Abdullah Y. Muaad
    Shaina Raza
    Usman Naseem
    Hanumanthappa J. Jayappa Davanagere
    Applied Intelligence, 2023, 53 : 29845 - 29862
  • [29] Granular Computing for Text Mining: New Research Challenges and Opportunities
    Jing, Liping
    Lau, Raymond Y. K.
    ROUGH SETS, FUZZY SETS, DATA MINING AND GRANULAR COMPUTING, PROCEEDINGS, 2009, 5908 : 478 - +
  • [30] Challenges and Opportunities of Text-Based Emotion Detection: A Survey
    Al Maruf, Abdullah
    Khanam, Fahima
    Haque, Md. Mahmudul
    Jiyad, Zakaria Masud
    Mridha, M. F.
    Aung, Zeyar
    IEEE ACCESS, 2024, 12 : 18416 - 18450