Data

Each of the acquisition sessions contains a sentence of transcript text (variable content, but not fully free-text). The raw data acquired consist in the timestamp of the instant a key is pressed, the timestamp of the instant the key is released, and the key ASCII code. From this raw information, it is possible to extract more complex features. 

The data are arranged in four datasets:

1) Development set: 115'120 users provided in a single .npy file that contains a nested Python nested dictionary (user IDs: session IDs: data)

2) Evaluation set: data from 15'000 users, provided in a single .npy file that contains a shallow Python dictionary (anonymized sessions IDs: data)

3) Development set: 40'639 users provided in a single .npy file that contains a nested Python dictionary (user IDs: session IDs: data)

4) Evaluation set: data from 5'000 users, provided in a single .npy file that contains a shallow Python dictionary (anonymized sessions IDs: data)

Two text files ('desktop_comparison_list.txt', 'mobile_comparison_list.txt') are associated to each evaluation set. They contain the list of pairwise comparisons (as anonymized sessions IDs) to be carried out to generate the list of scores to be submitted.


The demographic labels of the development sets are included with the data.

Important Update (19/06/2023) 

Please note that the datasets provided are obtained from two public databases of mobile interaction data (desktop, and mobile) collected by the User Interfaces group of the Aalto University. 

In order to provide a fair evaluation, in the CodaLab platform we have considered two scenarios:

1) Constrained scenario: it is mandatory to use only the provided development dataset.

2) Unconstrained scenario: it is possible to use any data, as there are no restrictions. This scenario is thought especially for commercial systems, but anyone can participate.

In the case of the unconstrained scenario, the text files to be provided must be named 'desktop_unconstrained_predictions.txt', 'mobile_unconstrained_predictions.txt'.

In case of doubt, please write to giuseppe.stragapede@uam.es.