Identification Of Users with Social Behavioural Biometrics On Twitter

Online Social Network (OSN) platforms provide a valuable source of information, which includes user profiles and social connections that can be used to develop behavioural models of individuals. Tweets of four Twitter users are analysed and a method is proposed to identify users. The Twitter text data is pre-processed to get meaningful words to extract features of every user. Term Frequency-Inverse Document Frequency (TF-IDF) is employed to eliminate features with lower weights, extract essential features from the text, and retrieve the corresponding word vectors using a pre-trained Word2Vec model. These word vectors are input to Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM) and CNN-LSTM for classification to generate a model to predict the user. The experimental results achieved an accuracy of 64.41% in identifying a user in a particular category. When the behavioural patterns of users change over time, it could be challenging to develop an accurate model calling for frequent updating of models. The proposed model can identify users on social media platforms by analyzing behavioural patterns from text data. This can enhance the security and trustworthiness of social media platforms. It can also identify attempts at impersonating other users by analysing behavioural patterns.