Text based personality prediction from multiple social media data sources using pre-trained language model and model averaging

Abstract

The ever-increasing social media users has dramatically contributed to significant growth as far as the volume of online information is concerned. Often, the contents that these users put in social media can give valuable insights on their personalities (e.g., in terms of predicting job satisfaction, specific preferences, as well as the success of professional and romantic relationship) and getting it without the hassle of taking formal personality test. Termed personality prediction, the process involves extracting the digital content into features and mapping it according to a personality model. Owing to its simplicity and proven capability, a well-known personality model, called the big five personality traits, has often been adopted in the literature as the de facto standard for personality assessment. To date, there are many algorithms that can be used to extract embedded contextualized word from textual data for personality prediction system; some of them are based on ensembled model and deep learning. Although useful, existing algorithms such as RNN and LSTM suffers from the following limitations. Firstly, these algorithms take a long time to train the model owing to its sequential inputs. Secondly, these algorithms also lack the ability to capture the true (semantic) meaning of words; therefore, the context is slightly lost. To address these aforementioned limitations, this paper introduces a new prediction using multi model deep learning architecture combined with multiple pre-trained language model such as BERT, RoBERTa, and XLNet as features extraction method on social media data sources. Finally, the system takes the decision based on model averaging to make prediction. Unlike earlier work which adopts a single social media data with open and close vocabulary extraction method, the proposed work uses multiple social media data sources namely Facebook and Twitter and produce a predictive model for each trait using bidirectional context feature combine with extraction method. Our experience with the proposed work has been encouraging as it has outperformed similar existing works in the literature. More precisely, our results achieve a maximum accuracy of 86.2% and 0.912 f1 measure score on the Facebook dataset; 88.5% accuracy and 0.882 f1 measure score on the Twitter dataset.

Knowledge Graph-Enabled Text-Based Automatic Personality Prediction

“…Moreover, they have concatenated the deep semantic features with the statistical linguistic features obtained directly from the text posts, and have fed them into a regression model to predict the Big Five personality traits' labels. Exploiting the embedding methods abilities, in their study Christian et al [17] have suggested a multi model deep learning architecture for personality prediction which was combined with various pre-trained language model including BERT, RoBERTa, and XLNet as a feature extraction method on social media text. The main idea behind their investigations was that, since the common deep learning models such as recurrent neural networks (RNNs) and LSTMs suffer from some drawbacks that are defeated using embedding methods, the embedding methods practically outperform them.…”

“…In the last few years, we have witnessed a considerable rise in text-based APP, that have used embedding methods to transfer the text elements to a more meaningful space (rather than character space), in favor of a better exploitation of computational methods [14][15][16][17].…”

Text-Based Automatic Personality Prediction Using KGrAt-Net; A Knowledge Graph Attention Network Classifier

“…A wide diversity of contributions has been published on deep learning-based APP, each of which has a distinct methodology. During last years, an increasing number of studies have investigated the application of embedding methods to transfer the text elements from a textual space to a real valued vector space [14][15][16]18]. The authors in [14], integrated traditional psycholinguistic features such as Mairesse, SenticNet, NRC Emotion Intensity Lexicon, and VAD Lexicon (a lexicon of over 20,000 English words annotated with their valence, arousal and dominance scores), with several language model embeddings, including Bidirectional Encoder Representation from Transformers (BERT), ALBERT (A Lite Biomedical BERT) and RoBERTa (A Robustly Biomedical BERT Approach) to predict personality from the Essays Dataset in Big Five model.…”

“…At last, they have used three different models (CNN, GRU and LSTM) to perform final predictions on MBTI and Big Five model. In an attempt to predict users' personality in social media, Christian et al [16] proposed a multi model deep learning system based on multiple pre-trained language model, including: BERT, RoBERTa, and XLNet as the features extraction methods. Similarly, they also believed that the existing algorithms suffers from the absence of context information.…”

Knowledge Graph-Enabled Text-Based Automatic Personality Prediction

“…Over the years, it has received much attention; while some studies have applied a combination of linguistic features and machine learning methods [ 6 – 9 ], some others have focused solely on the machine (deep) learning methods [ 10 – 13 ]. In the last few years, we have witnessed a considerable rise in text-based APP, which have used embedding methods to transfer the text elements to a more meaningful space (rather than character space), in favor of better exploitation of computational methods [ 14 – 17 ].…”