What Counts as Zero-Shot? A Systematic Review of Target-Language Data Use in Structured ABSA Transfer

Abstract In structured cross-lingual aspect-based sentiment analysis (ABSA), the label zero-shot can leave the scope of target-language data access unspecified. We examined two restrictions, one on human-annotated target-language task examples and one on any task-specific target-language material beyond the current evaluation input. A structured search of the Association for Computational Linguistics (ACL) Anthology on 15 November 2026 and supplementary discovery yielded 114 unique bibliographic records. Of 38 prioritized reports, 20 had provisional eligibility support, nine were contextual and nine remained unresolved. Independent screening and reconciliation are pending. We traced resource provenance, supervision and stage of access in 16 selected configurations from nine reports. The evidence supported absence on both checks in four configurations. For six, it supported absence of human-annotated target examples and documented other target-language material. For four, it documented target material and left use of human annotations unresolved. Both checks were unresolved in two. Five within-study contrasts documented score differences under aligned evaluations. Resource access could change alongside training volume or inference operations, and different scoring protocols limited comparison across studies. For each experimental arm, we propose specifying the resources withheld and the stages covered by the zero-shot claim, and linking that claim to the data, selection procedure and scorer.

Authors

Institutions

Publication Details

Journal
Research Square
Published
2026-11-17
DOI
https://doi.org/10.21203/rs.3.rs-11746069/v1
Primary Topic
Sentiment Analysis and Opinion Mining
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

What Counts as Zero-Shot? A Systematic Review of Target-Language Data Use in Structured ABSA Transfer

Elov Botir Boltayevich, Xusainova Zilola Yuldashevna, Ruhillo Alaev
2 links
Research Square
Sentiment Analysis and Opinion Mining
preprint

What Counts as Zero-Shot? A Systematic Review of Target-Language Data Use in Structured ABSA Transfer

Elov Botir Boltayevich, Xusainova Zilola Yuldashevna, Ruhillo Alaev
preprint en

Abstract

Abstract In structured cross-lingual aspect-based sentiment analysis (ABSA), the label zero-shot can leave the scope of target-language data access unspecified. We examined two restrictions, one on human-annotated target-language task examples and one on any task-specific target-language material beyond the current evaluation input. A structured search of the Association for Computational Linguistics (ACL) Anthology on 15 November 2026 and supplementary discovery yielded 114 unique bibliographic records. Of 38 prioritized reports, 20 had provisional eligibility support, nine were contextual and nine remained unresolved. Independent screening and reconciliation are pending. We traced resource provenance, supervision and stage of access in 16 selected configurations from nine reports. The evidence supported absence on both checks in four configurations. For six, it supported absence of human-annotated target examples and documented other target-language material. For four, it documented target material and left use of human annotations unresolved. Both checks were unresolved in two. Five within-study contrasts documented score differences under aligned evaluations. Resource access could change alongside training volume or inference operations, and different scoring protocols limited comparison across studies. For each experimental arm, we propose specifying the resources withheld and the stages covered by the zero-shot claim, and linking that claim to the data, selection procedure and scorer.

Research Square
Alisher Navo'i Tashkent State University of Uzbek Language and Literature (UZ), National University of Uzbekistan (UZ)
Sentiment Analysis and Opinion Mining
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon