CoChat
Sign Up
Cited by 3

The DOI matches its Crossref record.

Effects of Sampling on Twitter Trend Detection

Yates, Andrew, Kolcz, Alek, Goharian, Nazli, Frieder, Ophir·2016·Proceedings of the Language Resources and Evaluation Conference
DOI 10.63317/496yiwq6afa6
Open this paper in CoChat — free account

Overview

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.63317/496yiwq6afa6

Cites

Statistics

Figures

Figure 1 Flu prevalence estimates from Twitter's full feed of public tweets using an unsupervised method. Even when using an unsupervised method, the Twitter prevalence estimates are strongly correlated with CDC data and Google Flu Trends estimates, with Spearman's ρ = 0 . 79 for Google Flu Trends and ρ = 0 . 73 for the CDC data.
Figure 1 Flu prevalence estimates from Twitter's full feed of public tweets using an unsupervised method. Even when using an unsupervised method, the Twitter prevalence estimates are strongly correlated with CDC data and Google Flu Trends estimates, with Spearman's ρ = 0 . 79 for Google Flu Trends and ρ = 0 . 73 for the CDC data.CC-BY-NC · From the PDF
Figure 2 Correlation with Google Flu Trends for various cities as the Twitter sample size increases from a 1% sample to the full feed. Every city's correlation sharply increases until the sampling level reaches about 5%, at which point the correlations begin to increase slowly.
Figure 2 Correlation with Google Flu Trends for various cities as the Twitter sample size increases from a 1% sample to the full feed. Every city's correlation sharply increases until the sampling level reaches about 5%, at which point the correlations begin to increase slowly.CC-BY-NC · From the PDF
Image 3
CC-BY-NC · From the PDF
Image 4
CC-BY-NC · From the PDF
Figure 5 Correlation with Google Flu Trends for various cities when aggregating tweets at the month level. As with week-level aggregation, each city's correlation sharply increases until the sampling level reaches approximately 5%. The rate of increase is lower than that observed with weeklevel aggregation, however.
Figure 5 Correlation with Google Flu Trends for various cities when aggregating tweets at the month level. As with week-level aggregation, each city's correlation sharply increases until the sampling level reaches approximately 5%. The rate of increase is lower than that observed with weeklevel aggregation, however.CC-BY-NC · From the PDF
Figure 6 Correlation with Google Flu Trends as the fraction of the city's maximum correlation. While the correlation sharply increases with both aggregation levels, the increase is sharper with monthly aggregation. Monthly aggregation reaches 90% of its maximum correlation once the sampling level reaches about 6%, whereas weekly aggregation does not reach 90% of its maximum correlation until the sampling level reaches about 13%.
Figure 6 Correlation with Google Flu Trends as the fraction of the city's maximum correlation. While the correlation sharply increases with both aggregation levels, the increase is sharper with monthly aggregation. Monthly aggregation reaches 90% of its maximum correlation once the sampling level reaches about 6%, whereas weekly aggregation does not reach 90% of its maximum correlation until the sampling level reaches about 13%.CC-BY-NC · From the PDF
Figure 7 The relationship at the 1% sampling level between the number of flu tweets in a city and the Pearson correlation between Twitter flu prevalence estimates and Google Flu Trends' estimates. The tweet count and GFT correlation are fairly correlated, with Pearson's r = 0 . 25 and Spearman's ρ = 0 . 69 , suggesting a strong monotonic relationship and weak linear relationship.
Figure 7 The relationship at the 1% sampling level between the number of flu tweets in a city and the Pearson correlation between Twitter flu prevalence estimates and Google Flu Trends' estimates. The tweet count and GFT correlation are fairly correlated, with Pearson's r = 0 . 25 and Spearman's ρ = 0 . 69 , suggesting a strong monotonic relationship and weak linear relationship.CC-BY-NC · From the PDF
Figure 9 The relationship at the 1% sampling level between the number of flu tweets in a city and the Spearman correlation between Twitter flu prevalence estimates and Google Flu Trends' estimates. The tweet count and GFT correlation are relatively highly correlated, with Pearson's r = 0 . 37 and Spearman's ρ = 0 . 86 .
Figure 9 The relationship at the 1% sampling level between the number of flu tweets in a city and the Spearman correlation between Twitter flu prevalence estimates and Google Flu Trends' estimates. The tweet count and GFT correlation are relatively highly correlated, with Pearson's r = 0 . 37 and Spearman's ρ = 0 . 86 .CC-BY-NC · From the PDF
Figure 8 The relationship at the 10% sampling level between the number of flu tweets in a city and the Pearson correlation between Twitter flu prevalence estimates and Google Flu Trends' estimates. The tweet count and GFT correlation are much less correlated than at the 1% sampling level, with Pearson's r = 0 . 12 and Spearman's ρ = 0 . 45 .
Figure 8 The relationship at the 10% sampling level between the number of flu tweets in a city and the Pearson correlation between Twitter flu prevalence estimates and Google Flu Trends' estimates. The tweet count and GFT correlation are much less correlated than at the 1% sampling level, with Pearson's r = 0 . 12 and Spearman's ρ = 0 . 45 .CC-BY-NC · From the PDF
Figure 10 The relationship at the 10% sampling level between the number of flu tweets in a city and the Spearman correlation between Twitter flu prevalence estimates and Google Flu Trends' estimates. The tweet count and GFT correlation are fairly correlated, with Pearson's r = 0 . 26 and Spearman's ρ = 0 . 69 .
Figure 10 The relationship at the 10% sampling level between the number of flu tweets in a city and the Spearman correlation between Twitter flu prevalence estimates and Google Flu Trends' estimates. The tweet count and GFT correlation are fairly correlated, with Pearson's r = 0 . 26 and Spearman's ρ = 0 . 69 .CC-BY-NC · From the PDF
Figure 11 The relationship between the number of flu tweets in a city and the fraction of the maximum GFT Pearson correlation achieved at the 1% sampling level (i.e., corr 0 . 01 /corr 1 . 0 ). There is a weak Pearson correlation ( r = 0 . 23 ) and strong Spearman correlation ( ρ = 0 . 75 ) between the two.
Figure 11 The relationship between the number of flu tweets in a city and the fraction of the maximum GFT Pearson correlation achieved at the 1% sampling level (i.e., corr 0 . 01 /corr 1 . 0 ). There is a weak Pearson correlation ( r = 0 . 23 ) and strong Spearman correlation ( ρ = 0 . 75 ) between the two.CC-BY-NC · From the PDF
Figure 12 The correlation between the tweet count and GFT Pearson correlation at each sampling level. The correlation decreases sharply as the sampling level increases, with Pearson's r = 0 . 25 at the 1% sampling level decreasing to r = 0 . 12 at the 10% sampling level before reaching its final value of r = 0 . 07 at the 100% sampling level.
Figure 12 The correlation between the tweet count and GFT Pearson correlation at each sampling level. The correlation decreases sharply as the sampling level increases, with Pearson's r = 0 . 25 at the 1% sampling level decreasing to r = 0 . 12 at the 10% sampling level before reaching its final value of r = 0 . 07 at the 100% sampling level.CC-BY-NC · From the PDF

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

No references could be resolved yet.

Research works printed but not linked: 24

  1. Achrekar, H., Gandhe, A., Lazarus, R., Yu, S.-H., and Liu, B. (2012). Twitter improves seasonal influenza predic- tion. In Emmanuel Conchon, et al., editors, HEALTH- INF, pages 61-70.
  2. Aramaki, E., Maskawa, S., and Morita, M. (2011). Twit- ter catches the flu: Detecting influenza epidemics using twitter. In Proceedings of the Conference on Empirical Methods in Natural Language Processing, EMNLP '11, pages 1568-1576, Stroudsburg, PA, USA. Association for Computational Linguistics.
  3. Chakraborty, P., Khadivi, P., Lewis, B., Mahendiran, A., Chen, J., Butler, P., Nsoesie, E. O., Mekaru, S. R., Brownstein, J. S., Marathe, M., et al. (2014). Forecast- ing a moving target: Ensemble models for ili case count predictions. In SIAM international conference on Data Mining.
  4. Chew, C. and Eysenbach, G. (2010). Pandemics in the age of twitter: Content analysis of tweets during the 2009 h1n1 outbreak. PLoS ONE, 5(11):1-13, 11.
  5. Culotta, A. (2010). Towards detecting influenza epidemics by analyzing twitter messages. In Proceedings of the First Workshop on Social Media Analytics, SOMA '10, pages 115-122, New York, NY, USA. ACM.
  6. de Quincey, E. and Kostkova, P. (2010). Early warning and outbreak detection using social networking web- sites: The potential of twitter. In Patty Kostkova, edi- tor, Electronic Healthcare, volume 27 of Lecture Notes of the Institute for Computer Sciences, Social Informat- ics and Telecommunications Engineering, pages 21-24. Springer Berlin Heidelberg.
  7. Gesualdo, F., Stilo, G., Agricola, E., Gonfiantini, M. V., Pandolfi, E., Velardi, P., and Tozzi, A. E. (2013). Influenza-like illness surveillance on twitter through au- tomated learning of nave language. PLoS ONE, 8(12), 12.
  8. Ghosh, S., Zafar, M. B., Bhattacharya, P., Sharma, N., Gan- guly, N., and Gummadi, K. (2013). On sampling the wisdom of crowds: Random vs. expert sampling of the twitter stream. In Proceedings of the 22Nd ACM Inter- national Conference on Conference on Information & Knowledge Management, CIKM '13, pages 1739-1744, New York, NY, USA. ACM.
  9. Ginsberg, J., Mohebbi, M., Patel, R., Brammer, L., Smolin- ski, M., and Brilliant, L. (2009). Detecting influenza epidemics using search engine query data. Nature, 457:1012-1014. doi:10.1038/nature07634. · DOI · Paper page
  10. Ji, X., Chun, S., and Geller, J. (2012). Epidemic outbreak and spread detection system based on twitter data. In Jing He, et al., editors, Health Information Science, vol- ume 7231 of Lecture Notes in Computer Science, pages 152-163. Springer Berlin Heidelberg.
  11. Joseph, K., Landwehr, P., and Carley, K. (2014). Two 1twitters streaming api. In WilliamG. Kennedy, et al., editors, Social Computing, Behavioral-Cultural Model- ing and Prediction, volume 8393 of Lecture Notes in Computer Science, pages 75-83. Springer International Publishing.
  12. Kergl, D., Roedler, R., and Seeber, S. (2014). On the en- dogenesis of twitter's spritzer and gardenhose sample streams. In Advances in Social Networks Analysis and Mining (ASONAM), 2014 IEEE/ACM International Con- ference on, pages 357-364, Aug.
  13. Lamb, A., Paul, M. J., and Dredze, M. (2013). Separating fact from fear: Tracking flu infections on twitter. Pro- ceedings of NAACL-HLT, pages 789-795.
  14. Lampos, V. and Cristianini, N. (2010). Tracking the flu pandemic by monitoring the social web. In Cognitive Information Processing (CIP), 2010 2nd International Workshop on, pages 411-416, June.
  15. Morstatter, F., Pfeffer, J., Liu, H., and Carley, K. (2013). Is the sample good enough? comparing data from twitter's streaming api with twitter's firehose. In International AAAI Conference on Weblogs and Social Media.
  16. Morstatter, F., Pfeffer, J., and Liu, H. (2014). When is it biased?: Assessing the representativeness of twitter's streaming api. In WWW (Companion Volume), pages 555-556.
  17. Nagar, R., Yuan, Q., Freifeld, C., Santillana, M., Nojima, A., Chunara, R., and Brownstein, J. (2014). A case study of the new york city 2012-2013 influenza season with daily geocoded twitter data from temporal and spa- tiotemporal perspectives. Journal of Medical Internet Research, 16(10):e236, 10.
  18. Nagel, A. C., Tsou, M.-H., Spitzberg, B. H., An, L., Gawron, J. M., Gupta, D. K., Yang, J.-A., Han, S., Peddecord, K. M., Lindsay, S., et al. (2013). The complex relationship of realspace events and messages in cyberspace: case study of influenza and pertussis using tweets. Journal of medical Internet research, 15(10):e237.
  19. Sadilek, A., Kautz, H. A., and Silenzio, V. (2012). Mod- eling spread of disease from social interactions. In Pro- ceedings of the Sixth International Conference on We- blogs and Social Media, Dublin, Ireland, June 4-7, 2012.
  20. Sampson, J., Morstatter, F., Maciejewski, R., and Liu, H. (2015). Surpassing the limit: Keyword clustering to im- prove twitter sample coverage. In Proceedings of the 26th ACM Conference on Hypertext & Social Media, HT '15, pages 237-245, New York, NY, USA. ACM.
  21. In Martin Szomszor et al., editors, Electronic Health- care, volume 69 of Lecture Notes of the Institute for Computer Sciences, Social Informatics and Telecom- munications Engineering, pages 18-26. Springer Berlin Heidelberg.
  22. Yates, A., Parker, J., Goharian, N., and Frieder, O. (2014). A framework for public health surveillance. In Proceed- ings of the Ninth International Conference on Language Resources and Evaluation (LREC'14). European Lan- guage Resources Association (ELRA).
  23. Yates, A., Goharian, N., and Frieder, O. (2016a). Learn- ing the relationships between drug, symptom, and medi- cal condition mentions in social media. In International AAAI Conference on Weblogs and Social Media.
  24. Yates, A., Joselow, J., and Goharian, N. (2016b). The news cycles influence on social media activity. In Interna- tional AAAI Conference on Weblogs and Social Media.

Cited by

Sorted by citations

See all 3

Metadata

Authors
Yates, A., Kolcz, A., Goharian, N., & Frieder, O.
Venue
Proceedings of the Language Resources and Evaluation Conference
Publisher
European Language Resources Association (ELRA)
Published
2016-05
Pages
2998-3005
ISSN
2522-2686
DOI
10.63317/496yiwq6afa6
Cited by
2
References
0

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Effects of Sampling on Twitter Trend Detection. https://app.cochat.ai/papers/10.63317/496yiwq6afa6
The paper
Yates, A., Kolcz, A., Goharian, N., & Frieder, O. (2016). Effects of Sampling on Twitter Trend Detection. Proceedings of the Language Resources and Evaluation Conference. https://doi.org/10.63317/496yiwq6afa6