Classifying and Extracting Data from Facebook Posts for Online Persona Identification
Hazel Brosas, Eugene Lim, Danica Sevilla, Denise Brentan Silva, Ethel Ong · Institutional Repositories DataBase (IRDB) · 2018
Large amount of user-generated data are posted online in social media platforms, including user preferences, dining and leisure activities, events, news and personal blogs.This resulted in varying efforts to process social media data using NLP and ML algorithms for topic classification, sentiment analysis and detection, and events classification.Such information are problematic to process, as they tend to be short, informal, inconsistent, and are highly contextualized.A series of tasks is involved from collecting, pre-processing, classification and extraction before social media data can be used.In this study, we built a multi-class classifier model to process Facebook posts in order to identify a user's online persona based on his/her preferences.Information extraction is then applied to find relevant data from the classified posts that can be used to generate a description of the user's online persona.The classifier currently achieves an accuracy of 76.02% and an F1 score of 73.10% using 10-fold cross validation from a dataset containing 16,682 posts.