Datasets previously analysed or used in collaborative research

External corpora NIPG has worked with, and what each one actually establishes. This is an experience record, not a claim of ownership or continued access.

External benchmark and interaction corpora

DatasetExperience demonstratedAppropriate wiki claim
ChaLearn First ImpressionsFirst-impression and perceived-personality modellingPreviously used external benchmark; identify the exact training or evaluation role from the relevant methods paper
MULTISIMOEvaluation against external-observer labels in a different data domainOut-of-distribution evaluation experience
UDIVADyadic interaction across different tasksAnalysis of task-dependent perceived personality
ELEASmall-group collaboration and group performanceGroup-level behavioural analysis
AMI Meeting CorpusA sequence of task-oriented meetingsAnalysis of changes across interactions and tasks
MeMoMemory and repeated social interactionJoint dataset publication and Delft collaboration

The original ChaLearn dataset is described in ChaLearn LAP 2016 First Round Challenge on First Impressions.

These corpora are the evidence base behind perceived personality, including the ELEA result on group performance.

MeMo in detail

Introducing MeMo: A Multimodal Dataset for Memory Modelling in Multiparty Conversations (Tsfasman, Dudzik, Fenech, Lőrincz, Jonker, Oertel — arXiv). 31 hours of small-group discussion, repeated three times over two weeks, with participant memory reports and multimodal behavioural and perceptual measures.

Safeguard. Not a clinical dataset.

The repeated-measurement structure is methodologically the closest external analogue to what the Speech Gap longitudinal study would produce.

Internal archived material

The 2013 educational-games recordings — 180 primary-school participants in the study, 57 analysed on video. Potentially reusable with current methods, pending a feasibility check on file and permission quality.

Safeguard. No raw video of children in the public wiki. Aggregate findings and dataset descriptions only.

→ Adaptive assessment and serious games

Prospective data

SourceStatus
Speech Gap / BPD longitudinal recordingsPlanned; protocol not yet registered
Danish registers and biobanksOpportunity only; no access established, and registers do not necessarily contain repeated speech recordings
Camera-based therapy monitoringOpportunity only; no agreed study, no project plan

Perceived personality · Supporting collaborations · Data governance and annotation · Bibliography