Finding Characteristics of Users in Sensory Information: From Activities to Personality Traits

Abstract

The main objective of this work was to use information provided by a sensor-based activity recognition system to create a profile comprising user habits and link this information to the personality traits of users. User habits are represented by the sequence and duration of often observed daily life activities. Based on this information, we represented the user trials (sequence of activities) following a numerical method using Fourier series. The duration and sequence of the activities changed the phase and amplitude of the harmonics present in the Fourier representation. Each trial represented in this manner is called a behavioral spectrum. These data and the scores obtained from personality questionnaires were clustered separately and then an association was created between the clusters. The objective was to associate the activity-related features (sensor-based) and personality traits. The experimental results showed that for both young and elderly subjects, there is an association between the user personality traits and the manner in which they perform their activities. Moreover, the results obtained in this work show a promising method of assessing the personality traits of users based on their activities.

Deep Sensing: Inertial and Ambient Sensing for Activity Context Recognition Using Deep Convolutional Neural Networks

“…Research activities in the human activity context can be broadly categorized into two. Video-based human activity recognition and sensor-based activity recognition [ 4 , 7 , 8 , 9 , 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 , 18 ]. The sensor-based activity recognition process focused more on using data generated by inertial sensors, such as accelerometer and gyroscope, for recognizing human locomotive activities by either placing these sensors on various parts of the human body or using smartphones [ 16 , 24 , 25 , 26 ].…”

“…Other works based on classical machine learning algorithms and handcrafted feature extraction processes have been extensively reported [ 4 , 24 , 25 , 26 , 27 ]. For example, authors in Reference [ 18 ] proposed a new approach using a descriptor-based approach to human activity recognition. They handcrafted time and frequency domain features from accelerometer and gyroscope signals and then used conventional support vector machines and k-nearest neighbor algorithms.…”

“…In recent years, several works also focused on using deep neural networks for activity recognition using signals from only inertial sensors [ 5 , 6 , 7 , 8 , 9 , 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 , 29 , 30 ]. This new development is due to the incredible advancements in compute power.…”

“…[ 6 , 7 , 8 , 9 , 10 , 11 ]. DCNNs are now widely used for the development of automatic human activity context recognition [ 6 , 7 , 8 , 9 , 12 , 13 , 14 , 15 , 16 , 17 , 18 ]. Representational learning of activity context from raw sensor data using a DCNN has been proposed for automatic feature extraction in activity context recognition [ 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 , 20 , 21 ].…”

Correlation based data unification for personality trait prediction

“…Dandannavar et al [11], investigated the influence on demographics and Big Five personality dimensions on social media. The word vector-based Bi-RNN model is developed for personality prediction based on word vector representation [12].…”