Open tools for exploring dynamic interaction and multimodal signals in social-interaction research.
A DIMS dashboard puts video, time series, transcripts and annotations in one place, against one shared clock, so that qualitative and quantitative readings of an interaction can sit beside each other — recurrence quantification, cross-recurrence, cross-wavelet coherence, a cross-effector network and ELAN annotations included. Coherence is read against a chance level estimated by simulation: two unrelated signals score about 0.25, not 0, so a number without that comparison cannot be interpreted.
The core is released as one version that everything else pins. The current one is v1.0.1.
🌐 dims-network.github.io · 📘 Documentation · 🚀 Getting started
One repository holds all the code. Each study is a small repository of its own, holding its configuration, its data, and a pinned copy of that code — verified against the release in CI, so a study can never quietly fork.
| dims | the core — dashboard, tabs, Python analyses, no-code builder | everything else pins a release of this |
| case-demo | the reference study, with real data — rebuilt on each release | live |
| dims-dashboard-demo | the same study, frozen at 1.0.0 and self-contained: it opens with no network and does not move | live |
| case-ortho | ORTHO — a two-person tabletop game | live |
| case-karnatak | Karnatak music lessons | private: the recordings are identifiable |
| dims-network.github.io | the site, generated from the core's markdown |
Separately, bots-viewer and Ortho-viewer are Dash applications for browsing many game sessions at once. They are not dashboards and do not pin the core.
The code used to live in five repositories kept in step by hand. They were not: three disjoint git lineages, four copies of the frontend between 2150 and 2514 lines, and no repository that contained every feature. One fork carried the only correct wavelet coherence for months while the others shipped a version that tracked signal power instead — because there was no mechanism for a fix to travel.
Propagation is now a version bump. The earlier repositories are archived and private: their history is kept, but they are closed rather than published, and old links into them will not resolve. Each has been superseded by one of the repositories above.
Much of this data is video of identifiable people. A study declares its
visibility once, in dims-case.json, and the declaration is enforced by
machinery rather than by memory: a commit hook, a push hook, and a CI check that
fails if restricted data is tracked — or if a study declaring itself private is
in a public repository. See
public and private data.
Issues labelled ready name the files, link the contract and state the
acceptance check, so one can be picked up without any other context. Start from
the core's README, a map
telling you which single document to read for the task you have.
Miao, G. Q., Trujillo, J., Bulls, L. S., Thornton, M. A., Dale, R., & Pouw, W. (2025). DIMS Dashboard for Exploring Dynamic Interactions and Multimodal Signals. Proceedings of the 47th Annual Meeting of the Cognitive Science Society (CogSci 2025).