Stephanie Fong

Monash University · Australia

Unclaimed · AI-structured profileOpenAlex

Monash University NLP researcher building benchmarks and datasets for safe, personalized mental-health LLMs.

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-17) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.

Research profile

Output—

Not yet scored: this profile lacks total publication-volume metrics beyond recent-3y counts. This axis would measure cumulative papers, venue tiers, and output trajectory once a fuller publication history accumulates.

Impact—

Not yet scorable: the profile shows h-index 1 and i10-index 0, and citations by year exist only for 2026 — the work is too recent for citation accumulation. This axis would measure citations, uptake, and influence as they accrue.

Focus100/100

The profile shows near-perfect topical concentration: five OpenAlex topics cluster around one theme — AI in healthcare and education (8) and XAI (8) lead, with AI ethics (6), ML in healthcare (4), and topic modeling (3) — and the field distribution lists a single field, NLP for mental-health conversational AI. All three Evolve catalog papers fall inside this lane.

Collaboration67/100

OpenAlex records 28 distinct coauthors against only 15 recent-3y works, and twelve recurring collaborators appear in the profile — including Yiwen Jiang, Zimu Wang, Zhongxing Xu, Zongyuan Ge, and Dominic Dwyer — indicating dense, team-based production. The mid-range score reflects a broad but still shallow network, and none of the three catalog papers lists Fong as corresponding author.

Activity81/100

Activity is one of the profile's strongest signals: countsByYear shows 15 works in 2026 alone, and recent3yWorks confirms 15 over the trailing three years — essentially all indexed output landed in a single, very recent year. Three 2026 arXiv preprints in the Evolve catalog corroborate the pace. The score is tempered only by the short observable window, which cannot yet demonstrate sustained multi-year productivity.

Reproducibility55/100

Reproducibility is scored from a thin base: reproMean is 55, computed over just one scored paper (reproPapers: 1). All three Evolve catalog papers carry a 'medium' readiness. Reviews consistently flag reproducibility-adjacent risks — fully LLM-synthetic sessions with GPT-4o judging in StratCBT, synthetic self-annotated data and moderate classifier F1 in SDARE-Bench, and a 128-sample gold set without significance tests in VIBE-Bench.

Research summary

Stephanie Fong is a natural language processing researcher at Monash University (Australia) whose work centers on evaluating and building large language model systems for mental-health and safety-critical conversation. The OpenAlex-resolved profile records 15 works in the last three years, all indexed in 2026, with an h-index of 1 and i10-index of 0 — an early-career output pattern consistent with recent entry into publishing. Topic frequencies concentrate on artificial intelligence in healthcare and education (8), explainable AI (8), ethics and social impacts of AI (6), machine learning in healthcare (4), and topic modeling (3), and the field distribution maps to mental-health conversational AI. Three 2026 arXiv papers appear in the Evolve catalog: SDARE-Bench, where Fong is first author and which offers the first benchmark for LLM stigma detection in dyadic and group dialogue; StratCBT, a 9.7K-session strategy-aligned CBT counseling dataset; and VIBE-Bench, probing personalized LLMs on cross-concept preference mapping. Evolve reviews praise the dataset-and-benchmark engineering while flagging caveats — synthetic data, GPT-4o judging, moderate classifier F1, and small gold sets without significance tests. Reproducibility averages 55 across one scored paper, with all three catalog entries rated medium readiness. The collaboration footprint spans 28 distinct coauthors.

Collaboration pattern

Fong works in a large, fluid team: 28 distinct coauthors across 15 recent-3y works, with twelve recurring names led by Yiwen Jiang, Zimu Wang, and Zhongxing Xu, alongside Zongyuan Ge and Dominic Dwyer. Authorship in the catalog is middle-heavy (two of three papers) with one first-author benchmark and no corresponding-author role, consistent with an early-career contributor embedded in Monash-centered NLP and health-AI groups.

Research keywords

mental health NLPLLM safety evaluationcognitive behavioral therapy dialoguebenchmark constructionpersonalized LLMspreference reasoningmulti-party conversationstigma detectionsynthetic dialogue generationLLM-as-judge evaluationsycophancyexplainable AI

Highlights

First-author benchmark for stigma in group dialogue
SDARE-Bench (2026, arXiv) is, per Evolve's review, the first benchmark targeting LLM stigma detection and response in dyadic and group dialogue; its group-pressure tests on sycophancy are striking, though synthetic self-annotated data and a moderate classifier F1 temper the conclusions.
9.7K-session strategy-aligned CBT counseling dataset
As a middle author on StratCBT (2026), Fong contributed to the first large strategy-aligned CBT dataset, spanning 9.7K counseling sessions; Evolve calls it valuable but notes it is fully LLM-synthetic and judged by GPT-4o, so realism claims need caution.
Where personalized LLMs fail at preference mapping
VIBE-Bench (2026) introduces PRCM and shows personalized LLMs failing cross-concept preference mapping across Big Five personality and Holland RIASEC profiles; Evolve finds the design compelling but notes the 128-sample gold set lacks significance tests.

Timeline

2026First-authored SDARE-Bench, the first benchmark for LLM stigma detection and response in dyadic and group dialogue (arXiv).
2026Co-authored StratCBT, a 9.7K-session strategy-aligned CBT counseling dataset (arXiv).
2026Co-authored VIBE-Bench, probing personalized LLMs on cross-concept preference mapping (arXiv).

Papers in the Evolve catalog

Frequent co-authors

Yiwen Jiang×3 Zimu Wang×2 Hongxi Yang×2 Xiangyu Zhao×2 Zhongxing Xu×2 Jiahe Liu×2 Zongyuan Ge×2 Dominic Dwyer×2 Yaling Shen×1 Hiu Weh Naomi Chow×1 Heung Ying Lai×1 Qingyang Xu×1

About this profile

This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.

Bibliometrics come from OpenAlex only (identity: openalex); h-index 1 and i10-index 0 reflect a very recently indexed profile, not a mature citation record.
Counts by year cover only 2026, so multi-year output and citation trends cannot be assessed.
All three catalog papers are 2026 arXiv preprints; peer-review status is not recorded.
No ORCID is provided; 'Stephanie Fong' is name-resolved from public bibliographic data and may overlap with same-name authors.