Title: A bootstrapping approach to social media quantification

Journal Article · · Social Network Analysis and Mining

Abstract This work considers the use of classifiers in a downstream aggregation task estimating class proportions, such as estimating the percentage of reviews for a movie with positive sentiment. We derive the bias and variance of the class proportion estimator when taking classification error into account to determine how to best trade off different error types when tuning a classifier for these tasks. Additionally, we propose a method for constructing confidence intervals that correctly adjusts for classification error when estimating these statistics. We conduct experiments on four document classification tasks comparing our methods to prior approaches across classifier thresholds, sample sizes, and label distributions. Prior approaches have focused on providing the most accurate point estimate while this work focuses on the creation of correct confidence intervals that appropriately account for classifier error. Compared to the prior approaches, our methods provide lower error and more accurate confidence intervals.

Research Organization:
Los Alamos National Laboratory (LANL), Los Alamos, NM (United States)
Sponsoring Organization:
USDOE; USDOE National Nuclear Security Administration (NNSA)
Grant/Contract Number:
89233218CNA000001
OSTI ID:
1812428
Report Number(s):
LA-UR--20-20309; 73; PII: 760
Journal Information:
Social Network Analysis and Mining, Journal Name: Social Network Analysis and Mining Journal Issue: 1 Vol. 11; ISSN 1869-5450
Publisher:
Springer Science + Business MediaCopyright Statement
Country of Publication:
Austria
Language:
English

References (21)

Measurement error in two-stage analyses, with application to air pollution epidemiology: MEASUREMENT ERROR IN TWO-STAGE ANALYSES journal December 2013
An Introduction to the Bootstrap book January 1993
Quantifying counts and costs via classification journal June 2008
From classification to quantification in tweet sentiment analysis journal April 2016
Learning from noisy label proportions for classifying online social data journal November 2017
Using ensembles for problems with characterizable changes in data distribution: A case study on quantification journal March 2017
On the study of nearest neighbor algorithms for prevalence estimation in binary problems journal February 2013
Bootstrapping probability-proportional-to-size samples via calibrated empirical population journal August 2013
Correcting for Exposure Measurement Error in a Reanalysis of lung cancer Mortality for the Colorado Plateau Uranium Miners Cohort journal January 1999
Supervised Learning by Training on Aggregate Outputs conference October 2007
Quantification via Probability Estimators
  • Bella, Antonio; Ferri, Cesar; Hernandez-Orallo, Jose
  • 2010 IEEE 10th International Conference on Data Mining (ICDM), 2010 IEEE International Conference on Data Mining https://doi.org/10.1109/ICDM.2010.75
conference December 2010
Quantification Trees
  • Milli, Letizia; Monreale, Anna; Rossetti, Giulio
  • 2013 IEEE International Conference on Data Mining (ICDM), 2013 IEEE 13th International Conference on Data Mining https://doi.org/10.1109/ICDM.2013.122
conference December 2013
Model-based bootstrapping when correcting for measurement error with application to logistic regression: Model-Based Bootstrapping journal May 2017
Tackling concept drift by temporal inductive transfer conference January 2006
Quantification and semi-supervised classification methods for handling changes in class distribution conference January 2009
Characterizing debate performance via aggregated twitter sentiment conference January 2010
Towards detecting influenza epidemics by analyzing Twitter messages conference January 2010
Tweet Sentiment: From Classification to Quantification
  • Gao, Wei; Sebastiani, Fabrizio
  • ASONAM '15: Advances in Social Networks Analysis and Mining 2015, Proceedings of the 2015 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining 2015 https://doi.org/10.1145/2808797.2809327
conference August 2015
Ordinal Text Quantification
  • Da San Martino, Giovanni; Gao, Wei; Sebastiani, Fabrizio
  • SIGIR '16: The 39th International ACM SIGIR conference on research and development in Information Retrieval, Proceedings of the 39th International ACM SIGIR conference on Research and Development in Information Retrieval https://doi.org/10.1145/2911451.2914749
conference July 2016
A Review on Quantification Learning journal November 2017
SemEval-2016 Task 4: Sentiment Analysis in Twitter conference January 2016