Review of Inferring Latent Attributes from Twitter

TL;DR

This review summarizes methods from 2011-2013 for inferring latent attributes such as gender and political leaning from Twitter data, highlighting its potential applications in business and legal contexts.

Contribution

It consolidates existing literature on inferring demographic and political attributes from Twitter, emphasizing future research directions.

Findings

Latent attributes can be inferred from user behavior and neighborhood data.

Twitter data enables demographic and political inference with practical applications.

The field has potential for expansion with new data analysis techniques.

Abstract

This paper reviews literature from 2011 to 2013 on how Latent attributes like gender, political leaning etc. can be inferred from a person's twitter and neighborhood data. Prediction of demographic data can bring value to businesses, can prove instrumental in legal investigation. Moreover, political leanings can be inferred from the wide variety of user data available on-line. The motive of this review is to understand how large data sets can be made from available twitter data. The tweeting and re tweeting behavior of a user can be user to infer attributes like, gender, age etc. We explore in this text how this field can be expanded in future and possible avenues for future research.