English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Zendy Plus

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Zendy Tools

Zendy Open

Eliciting feedback from end users of NLP models can be beneficial forimproving models. However, how should we present model responses to users sothey are most amenable to be corrected from user feedback? Further, whatproperties do users value to understand and trust responses? We answer thesequestions by analyzing the effect of rationales (or explanations) generated byQA models to support their answers. We specifically consider decomposed QAmodels that first extract an intermediate rationale based on a context and aquestion and then use solely this rationale to answer the question. A rationaleoutlines the approach followed by the model to answer the question. Our workconsiders various formats of these rationales that vary according towell-defined properties of interest. We sample rationales from language modelsusing few-shot prompting for two datasets, and then perform two user studies.First, we present users with incorrect answers and corresponding rationales invarious formats and ask them to provide natural language feedback to revise therationale. We then measure the effectiveness of this feedback in patching theserationales through in-context learning. The second study evaluates how welldifferent rationale formats enable users to understand and trust model answers,when they are correct. We find that rationale formats significantly affect howeasy it is (1) for users to give feedback for rationales, and (2) for models tosubsequently execute this feedback. In addition, formats with attributions tothe context and in-depth reasoning significantly enhance user-reportedunderstanding and trust of model outputs.

What if you said that differently?: How Explanation Formats Affect Human Feedback Efficacy and User Perception