We employ one of the challenges of The Twelfth International Conference on Document Analysis and Recognition (ICDAR 2013). Predict if a handwritten document has been produced by a male or a female writer. The detail can be found at http://handwriting. qu.edu.qa/icdar-2013-competitions/ gender-prediction/. Besides being an interesting research topic of its own, automatically identifying subcateg...

TASKS 1. Given the first 1000 words in 1700 blogs, develop age (13-17, 23-27, 33-47) and gender (Male, Female) prediction algorithms, with a goal of beating a baseline prediction accuracies of 73.24% (for gender) and 70.23% (for age). 2. Given 488 images, develop age (under 30, 35-50, over 55) and gender (Male, Female) prediction algorithms, with a goal of beating a baseline prediction accuraci...

In this paper, we explore the task of gender classification using limited network data with an application to Fotolog. We take a heuristic approach to automating gender inference based on username, followers and network structure. We test our approach on a subset of 100,000 nodes and analyze our results to find that there is a lot of value in these limited information and that there is great pr...

The aim of this paper is to investigate the feasibility of predicting the gender of a text document’s author using linguistic evidence. For this purpose, termand style-based classification techniques are evaluated over a large collection of chat messages. Prediction accuracies up to 84.2% are achieved, illustrating the applicability of these techniques to gender prediction. Moreover, the revers...

We compare the performance of several automatic classification systems across a collection of different feature sets in detecting the gender of Japanese authors. The Japanese language is notable for the distinctiveness of the different manners and modes of speech used by the two genders--speakers of one gender often use different verb forms, sentence-final particles, and even personal pronouns ...

In this paper, we describe how we address the ICML 2004 Physiological Data Modeling Contest. For the gender prediction task, we employ 5 off-the-shelf machine learning methods: decision tree, neural networks, naive bayes, logistic regression, and Support Vector Machines. We use neural networks for the context prediction tasks. Most of the methods perform reasonably well, acknowledging the succe...

This paper proposes and contrastively evaluates several novel approaches to utilizing annotator rationales to improve the prediction of user gender in social media for English and Spanish. Our methods outperform state-of-the-art systems for Twitter gender prediction, and yield up to 28% error reduction relative to an otherwise identical system and training data without the use of annotator rati...