Explaining software fault predictions to spreadsheet users

Mukhtar, Adil; Hofer, Birgit; Jannach, Dietmar; Wotawa, Franz

doi:10.1016/j.jss.2023.111676

Cited by 1 publication

(2 citation statements)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Figure 1 provides examples of rationales. As humans often explain their decisions through natural language, it is reasonable to assume that this type of explanation is suitable for explaining AI to non-technical users (Miller et al, 2017;Cambria et al, 2023;Mukhtar et al, 2023). Furthermore, rationales can adopt domain-specific jargon, tailoring the explanation to domain expert knowledge (e.g., medical practice Meldo et al, 2020).…”

Section: A Framework For Rationalesmentioning

confidence: 99%

“…Moreover, humans might prefer using abstractive rationales (e.g., a comment) over extractive rationales (e.g., highlighting words) to explain their decisions. Possible approaches to improving the human-friendliness of rationales may be combining high-granularity rationales like words into sentences (e.g., changing "awesome", "film" to "this is an awesome film") (Meldo et al, 2020;Mukhtar et al, 2023), or adding more contextual information justifying the model's behavior.…”

Section: Faithfulness and Plausibilitymentioning

confidence: 99%

See 1 more Smart Citation

Human-annotated rationales and explainable text classification: a survey

Herrewijnen,

Nguyen,

Bex

et al. 2024

Front. Artif. Intell.

View full text Add to dashboard Cite

Asking annotators to explain “why” they labeled an instance yields annotator rationales: natural language explanations that provide reasons for classifications. In this work, we survey the collection and use of annotator rationales. Human-annotated rationales can improve data quality and form a valuable resource for improving machine learning models. Moreover, human-annotated rationales can inspire the construction and evaluation of model-annotated rationales, which can play an important role in explainable artificial intelligence.

show abstract