The 2018 "Multilingual Author Profiling on SMS" Task

Colocated with Forum for Information Retrieval Evaluation (FIRE'18),
7th-9th December 2018, DAIICT, Gandhinagar, India.

The resutls for the task are now available Click here

Follow us on Twitter

Welcome to FIRE'18-MAPonSMS

The main aim of FIRE'18-MAPonSMS (Multilingual Author Profiling on SMS) shared task is to identify an author's gender and age (both individually and jointly) from multilingual (Roman Urdu and English) text (or SMS messages). Please click here to register.

Author Profiling

Authorship profiling is a task where the objective is the identification of author's demographic traits (age, gender, native language, etc.) by analyzing author's written text. It has potential applications in many fields including forensic, security, marketing, fake profiles identification on social networking sites, capturing sender of harassing messages, etc. The profile of an author can be either: i) monolingual or ii) multilingual. In the former case, the entire author profile is written in one language, while in the latter case a single author profile will contain text in two or more languages.
This world has become a global village due to the emergence of the Internet and smart devices. People are more intended to connect with others in a virtual world (the online world) instead of the real world and this thing has affected the society norms as well. Blogs, Tweets, YouTube, Facebook posts and many other social media applications and websites are used as a platform on which people share their thoughts, their ideas and somehow their routine. Also, these platforms are connecting people of different races, nationalities and indeed distinct origins (of native language). People use a common language (like English) for communication purpose, but somehow, they have an inclination to their native languages. Even social media platforms are allowing multilingual settings, e.g. Facebook, Blogs, and Twitter. Therefore, the multilingual text is growing day by day over the Internet. Consequently, research on multilingual text is attracting the attention of the research community.
The majority of the previous work has focused on mono-lingual text for author profiling (for example, PAN Author Profiling Competitions). However, the problem of author profiling on multilingual, particularly on SMS messages has not been thoroughly explored in the past. The FIRE'18-MAPonSMS presents an exciting opportunity to research community in developing and evaluating models for multilingual author profiling on SMS messages.

Task Description

Given a multilingual (Roman Urdu and English) SMS messages profile, your task is to determine its author’s gender and age (both individually and jointly).

Related Work

We refer you to:
  • Mehwish Fatima, Saba Anwar, Amna Naveed, Waqas Arshad, Rao Muhammad Adeel Nawab, Muntaha Iqbal and Alia Masood (2018), "Multilingual SMS-based Author Profiling: Data and Methods", Natural Language Engineering, Cambridge University Press (To appear).
  • Mehwish Fatima, Komal Hasan, Saba Anwar, Rao Muhammad Adeel Nawab (2017), "Multilingual author profiling on Facebook", Information Processing & Management, Elsevier, pp: 886 - 904, Vol: 53, Issue: 4, Standard: 0306-4573
  • Francisco Rangel, Paolo Rosso, Moshe Koppel, Efstatios Stamatatos, Giacomo Inches. Overview of the Author Profiling Task at PAN 2013. In: Forner P., Navigli R., Tufis D. (Eds.)Notebook Papers of CLEF 2013 LABs and Workshops., vol. 1179
  • Shlomo Argamon, Moshe Koppel, James W. Pennebaker and Jonathan Schler (2009), Automatically profiling the author of an anonymous text", Communications of the ACM 52 (2): 119–123.