Towards a Multimodal Analysis to Predict Mental Illness in Twitter Platform

Abstract

The rising number of people with mental illness has become a major concern all over the world. Many efforts have been done to improve the process of detection and surveillance of people with mental illness and one of them is through analysing users' activities on social media platforms. Social media contains multi-content of data such as text, image and social interactions log. By mining all the data, it could help to determine the current mental state of social media users and detect those who are suffering from mental illness. However, most of the prior researches on mental illness in social media have mainly focused on single-based content rather than multi-based content. Due to a rapid growth of multi-content data, considering on single-based content only will ignore the complementary information that offered by other modalities. In consequence, single-based content does not provide an insight knowledge on the phenomenon of interest. This study aims to propose a design of multimodal fusion model, with the exploitation of all the multi-content data including text, image and social interaction data. This multimodal fusion model is expected to produce an accurate result for the prediction of mental illness among social media users.

Personal popular Name Identification Through Twitter Data

“…The Importance of name on the user profile to facilitate people to find us. A lot of benefits can be explored by social networking data, such as quantifying the mobility of urban inhabitant [23], understanding mobility patterns [24], inferring individual lifestyle patterns [25], sensing urban land use for urban planning application [26], identifying the city center [27], estimating user location [28], urban population [29], students' group collaborative [30], privacy awareness [31], consumer behavior [32] and predict mental illness [33]. In general, prior studies have focused on the field text (user posted text) and geo-location (location check-in) as the criteria to make the measurement.…”