Towards User Personality Profiling from Multiple Social Networks

Abstract

The exponential growth of online social networks has inspired us to tackle the problem of individual user attributes inference from the Big Data perspective. It is well known that various social media networks exhibit different aspects of user interactions, and thus represent users from diverse points of view. In this preliminary study, we make the first step towards solving the significant problem of personality profiling from multiple social networks. Specifically, we tackle the task of relationship prediction, which is closely related to our desired problem. Experimental results show that the incorporation of multi-source data helps to achieve better prediction performance as compared to single-source baselines.

TweetFit: Fusing Multiple Social Media and Sensor Data for Wellness Profile Learning

“…In this work, we presented one of the first studies on individual wellness profiling from sensor and social media data, which was handled by training the "TweetFit" framework to infer BMI category and "BMI Trend" personal wellness attributes. To facilitate further research, we released the multi-source multimodal dataset (Farseev 2017), which can be used for research on: user profiling ; multi-view timeline analysis (Jain and Jalali 2014;Akbari et al 2016); and user identification across multiple social networks.…”

“…To preserve users' privacy, the dataset is released in the form of data representations (features) and anonymized multi-source user timelines, instead of the original user posts (Farseev 2017). In the dataset, users are well distributed in all BMI categories.…”

“…Here, an individual wellness profile involves personal user attributes ) such as demographics (age, gender, occupation, etc.) (Farseev et al 2015), Body Mass Index (BMI) category 1 , personality (Buraya et al 2017), or chronic disease tendency (Akbari et al 2016). In our study, we focus on two important personal wellness attributes -BMI category and "BMI Trend' (the direction of BMI fluctuation over time -Increase/Decrease).…”

“…To do so, we treat individual wellness profiling as a regularized multi-task learning (MTL) problem, where different data source combinations for each inference category are represented as MTL "tasks". To facilitate further research, we release our multi-source multimodal sensor-social dataset (Farseev 2017) for public use.…”

Do we behave differently on Twitter and Facebook: Multi-view social network user personality profiling for content recommendation

“…Later, Farseev and Chua ( 2017a ) extended the framework to leverage sensor data and multi-source multi-task learning for wellness profiling. Buraya et al ( 2017 ) proposed to solve the problem of relationship status inference by applying ‘out of the box” machine learning on early-fused data from Twitter, Instagram, Facebook , and Foursquare , achieving a significant 17% increase in performance compared to unimodal learning. Going further, Tsai et al ( 2019 ) proposed a factorization method to model the intra-modal and inter-modal relationships within multimodal data inputs, which proved to be important for the incorporation of multimodal data into user profiling, while Buraya et al ( 2018 ) instead leveraged the temporal component of the multimodal data, being the first to apply deep learning methods for multi-view personality profiling.…”