Data for: Measuring Concept Semantic Relatedness through Common Spatial Pattern Feature Extraction on EEG Signals

Published: 9 April 2018| Version 1 | DOI: 10.17632/shzz5kbsgy.1
Hiram Calvo, Jose Luis Paredes Paredes, Jesus Figueroa Nazuno


This files correspond to the EEG samples measured on 18 individuals while they watched 150 pairs of nouns. The task consisted in determining if these pairs were semantically related or not. Nouns were presented in 5 blocks constituting 30 pairs each in a counterbalanced form, that is, a word was presented as a primer in the case where there is semantic relationship, and then the same word is used as a target, but in this case there is no semantic relationship. The participants have an average of 27 years old, have at least a bachelor degree, and Spanish is their native language—the test was carried out in Spanish. Each one of the participants declared not currently being under any medical treatment, and not having consumed any alcoholic substance for at least 24 hours before the experiment. The distribution of samples is as follows: 7 Women: S2, S3, S4, S11, S13, S16, S18 11 Men: S1, S5, S6, S7, S8, S9, S10, S12, S14, S15, S17 An Emotiv EPOC device was used for EEG signals acquisition. This device has 14 electrodes positioned according to the system 10–20 [27] in the following regions: AF3, F7, F3, FC5, T7, P7, O1, O2, P8, T8, FC6, F4, F8 and AF4, as shown in Figure 5, with a sampling rate of 128 Hz. Sampling resolution is 16 bits, from which 14 are effective, and 2 are used to identify and discard noise. The sampling bandwidth was 0.2–45 Hz, with digital cut filters at 50 and 60 Hz. See the included file "dataset description.docx" for detailed information on these files.



Principal Component Analysis, Similarity Measure, Electroencephalography, Natural Language Semantics