首页 > AI前沿 > UO-FIE: Combining Exact-Label Supervision with Graded Utility for Factivity Inference

UO-FIE: Combining Exact-Label Supervision with Graded Utility for Factivity Inference

arXiv机器学习 2026-09-24 00:24 6 阅读 查看原文

The Factivity Inference Evaluation 2026 (FIE2026) classifies Chinese context-hypothesis pairs into nine ordered factivity intervals.

Its evaluation metric rewards both exact predictions and proximity to the correct interval, while 64.1% of the 566 training examples belong to a single class.

In preliminary experiments, several mDeBERTa classification models predominantly predict the dominant class, whereas a Huber-regression baseline produces more predictions near the correct interval but fewer exact matches.

We introduce Utility-Oriented Factivity Inference (UO-FIE)

We introduce Utility-Oriented Factivity Inference (UO-FIE), a parameter-efficient system that combines exact-label supervision with graded utility.

UO-FIE predicts a distribution over the nine classes and combines hard-label supervision, utility-based soft targets, scheduled class weights, and an ordinal loss.

Evaluation

We evaluate expected-utility decoding in controlled comparisons and use ordinal calibration selected on out-of-fold predictions for the submitted system.

Results

Based on Qwen3.5-9B with LoRA, UO-FIE ranks first in the fine-tuning track with a macro utility of 0.8316.

A separate prompt-based ensemble ranks third in the non-fine-tuning track with a macro utility of 0.8450.