Evaluation without ground truth in social media research

Reza Zafarani, Huan Liu

Research output: Contribution to journalArticle

28 Scopus citations

Abstract

Scientific research demands reproducible and independently verifiable findings. The challenges introduced by humans' lack of knowledge about the future are further compounded by yearning to understand why things happen on social media. Without surveying users on social media, the gap between personal understanding and reality cannot be gauged. Practitioners and researchers alike in various fields, including statistics, computer science, sociology, psychology, epidemiology, and ethology, have developed a range of methods social media researchers can borrow and tweak in their search for reproducible evaluation methods for social media research. Consider designing a method that predicts the most likely time users will check their email messages or the restaurant they will most likely choose for dinner using their checkins, or personally reported locations, in social media.

Original languageEnglish (US)
Pages (from-to)54-60
Number of pages7
JournalCommunications of the ACM
Volume58
Issue number6
DOIs
StatePublished - Jun 1 2015
Externally publishedYes

ASJC Scopus subject areas

  • Computer Science(all)

Fingerprint Dive into the research topics of 'Evaluation without ground truth in social media research'. Together they form a unique fingerprint.

Cite this