|
" Nowcasting user behaviour with social media and smart devices on a longitudinal basis : "
Tsakalidis, Adam
| Document Type
|
:
|
Latin Dissertation
|
| Record Number
|
:
|
1103538
|
| Doc. No
|
:
|
TLets801154
|
| Main Entry
|
:
|
Tsakalidis, Adam
|
| Title & Author
|
:
|
Nowcasting user behaviour with social media and smart devices on a longitudinal basis :\ Tsakalidis, Adam
|
| College
|
:
|
University of Warwick
|
| Date
|
:
|
2018
|
| student score
|
:
|
2018
|
| Degree
|
:
|
Ph.D.
|
| Abstract
|
:
|
The adoption of social media and smart devices by millions of users worldwide over the last decade has resulted in an unprecedented opportunity for NLP and social sciences. Users publish their thoughts and opinions on everyday issues through social media platforms, while they record their digital traces through their smart devices. Mining these rich resources offers new opportunities in sensing real-world events and indices (e.g., political preference, mental health indices) in a longitudinal fashion, either at the macro (population)-, or at the micro(user)-level. The current project aims at developing approaches to “nowcast" (predict the current state of) such indices at both levels of granularity. First, we build natural language resources for the static tasks of sentiment analysis, emotion disclosure and sarcasm detection over user-generated content. These are important for opinion monitoring on a large scale. Second, we propose a general approach that leverages textual data derived from generic social media streams to nowcast political indices at the macro-level. Third, we leverage temporally sensitive and asynchronous information to nowcast the political stance of social media users, at the micro-level using multiple kernel learning. We then focus further on the micro-level modelling, to account for heterogeneous data sources, such as information derived from users' smart phones, SMS and social media messages, to nowcast time-varying mental health indices of a small cohort of users on a longitudinal basis. Finally, we present the challenges faced when applying such micro-level approaches in a real-world setting and propose directions for future research.
|
| Subject
|
:
|
H Social Sciences (General)
|
|
|
:
|
Q Science (General)
|
| Added Entry
|
:
|
University of Warwick
|
| |