Early-career Fudan University researcher in LLM-based social simulation and LLM-agent evaluation of scientific forecasting
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
2
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-16) 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—
The output axis is not scored: indexed productivity data is insufficient. It would measure total publication volume, venue quality, and output consistency across the record; the sparse early footprint (h-index 2) cannot yet support a reliable score.
Impact—
The impact axis is not scored because citation evidence is insufficient. It would measure citation counts, i10-index, and uptake of the two 2026 arXiv preprints; both are recent postings, so meaningful impact signals have not yet accumulated.
Focus0/100
The focus axis scores 0, reflecting an even 50/50 field split between LLM-based social simulation and LLM-agent evaluation / research forecasting rather than concentration in one line. Indexed topic records span topic modeling (6), multi-agent systems and negotiation (4), sentiment analysis (3), dialogue systems (3), and social robot interaction (2). At this career stage the breadth reads as exploratory, but it currently prevents a single-track research identity.
Collaboration66/100
A collaboration score of 66 rests on 27 distinct co-authors recorded in OpenAlex, with 12 recurring collaborators including Yingqian Wu, Xinnong Zhang, Jiayu Lin, and Xinyi Mou. Both Evolve-reviewed 2026 papers place Liang in middle-author, non-corresponding positions, consistent with contributor roles inside larger team projects rather than solo leadership — a pattern typical of an early-stage researcher embedded in a multi-institution NLP and social-simulation group.
Activity64/100
Activity scores 64 on a steady early record: OpenAlex indexes 8 works in the recent three-year window, with counts by year of 2 (2023), 3 (2024), 2 (2025), and 3 (2026). The cadence shows no gap years and a slight upward trend into 2026, when both catalogued preprints — SocioVerse2 and RAP — appeared on arXiv. Output volume is modest but continuous across four consecutive years.
Reproducibility55/100
Reproducibility averages 55 across the 2 papers Evolve has reviewed, both rated medium readiness. The verdicts flag concrete limitations: SocioVerse2's validation repackages the authors' companion studies, and RAP relies on model-built instruments with unreleased code. Neither paper has yet released artifacts supporting independent replication, so the mid-range score reflects honest reporting rather than demonstrated reproducibility.
Research summary
Jingcong Liang is an early-career researcher at Fudan University (CN) whose OpenAlex-resolved record splits evenly between two directions: LLM-based social simulation within computational social science, and LLM-agent evaluation applied to research forecasting in science of science. The catalogued record centers on two 2026 arXiv preprints, both reviewed by Evolve. SocioVerse2 introduces a longitudinal, dynamic social-simulation framework under a human-AI co-evolutionary paradigm, evaluated across seven cases. RAP uses LLM agents to predict arXiv field-level attention shares, finding that every agent forecast trails an exponentially weighted moving average baseline while exposing target-conditioned evidence acquisition biases. In both papers Liang is a middle author without corresponding status; Evolve rates reproducibility readiness medium (mean 55 across 2 papers), citing model-built instruments and unreleased code in RAP and validation that repackages the authors' companion studies in SocioVerse2. The broader footprint — 27 distinct co-authors, 12 recurring collaborators, and topic clusters in topic modeling, multi-agent negotiation, sentiment analysis, dialogue systems, and human-robot interaction — marks an active contributor inside a large NLP team rather than a solo line. Activity is steady at 8 works over the recent three-year window (2, 3, 2, and 3 works from 2023 to 2026). Citation impact is thin (h-index 2, i10-index 1), so the output and impact axes remain unscored.
Collaboration pattern
Liang's OpenAlex record lists 27 distinct co-authors and 12 recurring collaborators, including Yingqian Wu, Xinnong Zhang, Jiayu Lin, Xinyi Mou, Siyuan Wang, and Hanjia Lyu. In both Evolve-reviewed 2026 papers Liang is a middle author without corresponding status, indicating contributor roles within large team efforts based at Fudan University spanning LLM agents, evaluation, and simulation. No solo-author or senior-author work appears in the current record.
Research keywords
Highlights
SocioVerse2: longitudinal LLM social simulation under human-AI co-evolution
RAP: LLM agents trail an EWMA baseline in forecasting research attention
One record, two lines: simulated societies and simulated scientists
The indexed field distribution splits 50/50 between LLM-based social simulation and LLM-agent evaluation for research forecasting, echoed by topic counts led by topic modeling (6) and multi-agent systems and negotiation (4). The two catalogued preprints pair a simulation framework with an agent-evaluation study of science itself.
Timeline
- 2023First works indexed in OpenAlex: 2 works recorded this year.
- 20243 works indexed, matching the highest annual count in the current record.
- 20252 works indexed.
- 2026SocioVerse2 posted to arXiv: a longitudinal social simulation framework under a human-AI co-evolutionary paradigm.
- 2026RAP posted to arXiv: LLM agents forecast arXiv research attention and reveal evidence acquisition biases.
Papers in the Evolve catalog
Frequent co-authors
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.
- Bibliometric indicators (27 co-authors, topic counts, per-year work counts, h-index 2, i10-index 1) derive from OpenAlex under identity 'openalex'; homonymous researchers cannot be fully excluded.
- Only 2 arXiv preprints (both 2026) carry Evolve reviews; no peer-reviewed venue assessment is included yet.
- The output and impact radar axes are unscored because the underlying data is insufficient.
- The reproducibility mean of 55 rests on only 2 papers rated 'medium' readiness.