Koustuv Saha

Output—

This axis would measure publication volume and productivity over time. The current OpenAlex snapshot lacks a year-by-year output series sufficient to score it, so no output score is given; the visible records cover only a single recent period rather than a full career trajectory.

Impact—

This axis would measure citations, h-index trajectory, and scholarly uptake of the work. The snapshot reports h-index 1 and i10-index 0 — too thin, and possibly lagged by name resolution, to support a fair impact score — so the axis is left unscored rather than estimated.

Focus0/100

The focus axis scores at the bottom of the scale, consistent with a field distribution split exactly one third each across HCI/computational social science, human-AI decision making, and HCI/social computing. Topic counts reinforce this breadth: ethics and social impacts of AI (6), innovative human-technology interaction (4), and topic modeling, information retrieval, and decision-making (2 each). The portfolio is deliberately interdisciplinary rather than concentrated in a single subfield.

Collaboration65/100

The collaboration dimension scores 65, backed by 25 distinct coauthors in the recent OpenAlex record. Authorship roles alternate: last author on the FinAInce trust experiment and PeerPen, middle author on the larger cross-platform AI-companion disruption team. This suggests a mix of leading small HCI studies and contributing to group projects with collaborators such as Eshwar Chandrasekharan, Talayeh Aledavood, Renwen Zhang, and Daniel Kang.

Activity83/100

This is the strongest dimension at 83. The OpenAlex record lists 16 works in the last three years, and the only year given in the counts-by-year series — 2026 — already shows 16 publications, all appearing as arXiv preprints in this catalog. Such a concentrated burst indicates a highly productive current period, though single-year coverage prevents judging longer-term cadence.

Reproducibility32/100

Reproducibility scores 32, the mean over three assessed papers. Evolve rates the preregistered FinAInce experiment (N=285) and the PeerPen interview study as low reproducibility readiness, and the Bayesian interrupted time-series companion-disruption analysis as medium. Methodologically rigorous designs — preregistration, hierarchical Bayesian modeling — coexist with artifacts that are hard to replicate, such as interview corpora and proprietary platform data.