DNA-Inspired Online Behavioral Modeling and Its Application to Spambot Detection
Open accessAbstract
A novel, simple, and effective approach to modeling online user behavior extracts and analyzes digital DNA sequences from user online actions and uses Twitter as a benchmark to test the proposal. Specifically, the model obtains an incisive and compact DNA-inspired characterization of user actions. Then, standard DNA analysis techniques discriminate between genuine and spambot accounts on Twitter. An experimental campaign supports the proposal, showing its effectiveness and viability. Although Twitter spambot detection is a specific use case on a specific social media platform, the proposed methodology is platform and technology agnostic, paving the way for diverse behavioral characterization tasks.
AI snapshot
Written from the paper’s full text.
TL;DR
The paper proposes modeling online user behavior as “digital DNA” sequences of action characters and applies DNA sequence analysis to detect Twitter spambots. It reports that DNA fingerprinting outperforms several state-of-the-art spambot detection approaches on two Twitter test sets.
Key findings
- Digital DNA sequences encode user actions as strings; tweet type DNA (A = simple tweet, T = reply, C = retweet) better discriminated Bot2 spambots than tweet content DNA, while both showed long shared substrings for spambot groups and little to no similarity for genuine accounts.
- In the LCS analysis, Bot1 had a group of 400 accounts with a common substring of length 774; in Test-set2, the suspicious group was 400 accounts with LCS around 1750, and in Test-set1 it was 983 accounts with LCS around 400.
- DNA fingerprinting achieved MCC = 0.952 for Test-set1 and MCC = 0.867 for Test-set2, with Precision 0.982/1.000, Recall 0.972/0.858, Specificity 0.981/1.000, Accuracy 0.976/0.929, and F-Measure 0.977/0.923.
- It outperformed Yang et al. (MCC 0.043/0.287), Miller et al. (MCC 0.059/−0.043), and Ahmed et al. (MCC 0.886/0.847) on the two test sets.
Methods
- Unsupervised DNA fingerprinting: collect Twitter timeline behavioral data, extract digital DNA sequences up to 3,200 characters, compare sequences via Longest Common Substring using a linear-time generalized suffix tree algorithm, and identify steep drops in LCS curves as candidate splitting points.
- Data: manually verified spambots and genuine accounts; 991 Bot1 accounts (50.05%) and 464 Bot2 accounts (89.29%) were certified as spambots, and 3,474 accounts that answered natural-language questions were certified as human, totaling 4,929 accounts. Bot1 retweeted an Italian political candidate; Bot2 spammed Amazon.com product URLs.
- Evaluation: compared against supervised Yang et al. and unsupervised Miller et al. and Ahmed et al. using Precision, Recall, Specificity, Accuracy, F-Measure, and MCC; genuine accounts were randomly undersampled to match spambot group sizes for LCS comparisons.
Limitations
- The paper states that detection used only one type of digital DNA, namely tweet type DNA, and suggests multiple DNA types could be combined in an ensemble or voting system as future work.
- The paper notes Twitter spambot detection is a specific use case on a specific social media, though it argues the methodology is platform and technology agnostic.
- Evident from the text: evaluation is limited to two spambot groups and Twitter timelines; splitting-point selection relies on visual/derivative analysis of LCS curves, and the approach’s reusability across other spambot behaviors is not empirically tested.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
0 of 8 cited sources checked · 0 supported · 0 flagged
- Findingown data§ B. Uncovering novel Twitter spambots · p. 5
The supervised approach by Yang et al. [4] was unable to accurately distinguish spambots from genuine accounts, as shown by many false negatives and low recall.
“The supervised approach by Yang et al. [4] proved unable to accurately distinguish spambots from genuine accounts, as demonstrated by the considerable number of false negatives ( FN ) and the resulting very low Recall .”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [4] [4] C. Yang, R. Harkreader, and G. Gu, 'Empirical evaluation and new design for fighting evolving twitter spammers,' Information Forensics and Security, IEEE Transactions on , vol. 8, no. 8, pp. 1280-1293, 2013. -
Not checked in the free run.
- Findingown data§ A. Characterizing spambot behavior · p. 3
In Figure 1(a), there exists a group of 400 accounts in Bot1 with a substring of length 774 in common.
“for example, in Figure 1(a) there exists a group of 400 accounts in Bot1 with a substring of length 774 in common.”
Quote in the paperNot a substantial claim - Findingreasoning§ B. Uncovering novel Twitter spambots · p. 5
The unsupervised approach in [8] provided unsatisfactory results.
“Moreover, also the unsupervised approach in [8] provided unsatisfactory results.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [8] [8] Z. Miller, B. Dickinson, W. Deitrick, W. Hu, and A. H. Wang, 'Twitter spammer detection using data stream clustering,' Information Sciences , vol. 260, pp. 64-73, 2014. -
Not checked in the free run.
- Findingown data§ V. CONCLUSION · p. 6
Results show that the proposal outperforms best-of-breed algorithms commonly employed for spambot detection.
“Results show that our proposal outperforms best-of-breed algorithms that are commonly employed for this task.”
Quote in the paperSubstantial claim - Findingown data§ A. Characterizing spambot behavior · p. 3
Spambots present higher similarity in their digital DNA than genuine accounts.
“From figures 1(a) and 1(b) it is clear that the LCS of both the groups of spambots are rather long even when the number of accounts grows. This is strikingly evident for Bot2 (spammers of Amazon.com products), see Figure 1(b). In contrast, genuine accounts show little to no similarity.”
- Also stated in: § B. Digital DNA as a proxy of user online behavior · p. 2 · plain
“We show how groups of spambots share common patterns, as opposed to groups of genuine accounts.”
Quote in the paperSubstantial claim - Also stated in: § B. Digital DNA as a proxy of user online behavior · p. 2 · plain
- Findingnone given§ III. INTRODUCING DIGITAL DNA FINGERPRINTING · p. 2
Similar phenomena of hard-to-detect automated accounts are widespread outside Italy.
“We found similar phenomena to be widespread also outside Italy.”
Quote in the paperNot a substantial claim - Interpretationreasoning§ B. Uncovering novel Twitter spambots · p. 4
The steeper and more pronounced the drop in the LCS curve, the more different are the two groups split by that drop.
“Furthermore, the steeper and the more pronounced the drop in the LCS curve, the more different are the two groups of accounts split by that drop.”
Quote in the paperSubstantial claim - Interpretationown data§ A. Characterizing spambot behavior · p. 3
Modeling behavior according to the type of tweets, rather than their content, is more effective for discriminating Bot2 spambots.
“These results suggest that modeling the behavior of Twitter accounts according to the type of tweets, rather than to their content, is more effective for discriminating Bot2 spambots.”
Quote in the paperSubstantial claim - Premisereasoning§ III. INTRODUCING DIGITAL DNA FINGERPRINTING · p. 2
The automated accounts used in the Rome 2014 mayoral election were extremely hard to distinguish from genuine ones.
“Surprisingly, these automated accounts were extremely hard to distinguish from genuine ones.”
- Also stated in: § III. INTRODUCING DIGITAL DNA FINGERPRINTING · p. 2 · hedged
“they showed a tweeting behavior apparently similar to that of genuine accounts”
Quote in the paperSubstantial claim - Also stated in: § III. INTRODUCING DIGITAL DNA FINGERPRINTING · p. 2 · hedged
- Interpretationnone given§ B. Uncovering novel Twitter spambots · p. 5
Novel social spambots tweet contents similar to that of genuine accounts.
“However, novel social spambots tweet contents similar to that of genuine accounts (e.g., retweets of genuine tweets and famous quotes).”
Quote in the paperSubstantial claim - Interpretationreasoning§ B. Uncovering novel Twitter spambots · p. 4
Multiple plateaux in a single LCS curve indicate multiple subgroups among the accounts.
“Note that it is possible to observe multiple plateaux in a single LCS curve, as in the case of Figure 2(a). This represents a situation where multiple (sub-)groups exist among the whole set of considered accounts.”
Quote in the paperSubstantial claim - Interpretationown data§ B. Uncovering novel Twitter spambots · p. 4
Plateaux in LCS curves are strictly related to homogeneous groups of highly similar accounts.
“In fact, such plateaux in LCS curves are strictly related to homogeneous groups of highly similar accounts.”
Quote in the paperSubstantial claim - Interpretationreasoning§ B. Uncovering novel Twitter spambots · p. 6
The DNA-inspired technique is flexible enough to highlight suspicious similarities among groups of accounts without focusing on specific characteristics.
“Instead, our DNA-inspired technique is flexible enough to highlight suspicious similarities among groups of accounts without focusing on specific characteristics.”
Quote in the paperSubstantial claim - Interpretationreasoning§ B. Uncovering novel Twitter spambots · p. 3
In a mixed group of bot and genuine accounts, almost only bot accounts will have long DNA substrings in common.
“In a group with mixed bot and genuine accounts, almost only bot accounts will have long DNA substrings in common.”
Quote in the paperSubstantial claim- The authors state this as an assumption: Similarity in digital DNA is a proxy for automation.
- Interpretationreasoning§ B. Uncovering novel Twitter spambots · p. 6
The approach in [9] might lack reusability across other groups of spambots with different behaviors.
“However, although achieving an overall good performance for the considered spambots, the approach in [9] might lack reusability across other groups of spambots with different behaviors, such as those perpetrating a follower fraud [5],[11].”
Quote in the paperSubstantial claim0 of 2 sources checked- cites [5] [5] M. Jiang, P. Cui, A. Beutel, C. Faloutsos, and S. Yang, 'Catching synchronized behaviors in large networks: A graph mining approach,' ACM Transactions on Knowledge Discovery from Data (TKDD) , 2016 (to appear). - source
Not checked in the free run.
- cites [11] [11] S. Cresci, R. Di Pietro, M. Petrocchi, A. Spognardi, and M. Tesconi, 'Fame for sale: Efficient detection of fake Twitter followers,' Decision Support Systems , vol. 80, pp. 56-71, 2015. Stefano Cresci is a Ph.D. student at University of Pisa, Information Engineering Dept., and a research fellow at IIT-CNR in Pisa, Italy. His research interests include social media mining and knowledge discovery. Email: s.cresci@iit.cnr.it Dr. Roberto Di Pietro is with the Nokia Bell Labs, Paris, France. His research interests include security, privacy, distributed systems, computer forensics, and analytic source
Not checked in the free run.
- Premisecitation§ B. Uncovering novel Twitter spambots · p. 5
Among the 126 features proposed in [8], 95 are based on the textual content of tweets.
“Among the 126 features proposed in [8], 95 are based on the textual content of tweets.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [8] [8] Z. Miller, B. Dickinson, W. Deitrick, W. Hu, and A. H. Wang, 'Twitter spammer detection using data stream clustering,' Information Sciences , vol. 260, pp. 64-73, 2014. -
Not checked in the free run.
- May combine two claims. 9
- Classified as probably not a substantial claim. 5
May combine two claims.
Findingown data§ DNA-inspired online behavioral modeling and its application to spambot detection · p. 1The experimental campaign supports the DNA-inspired proposal, showing its effectiveness and viability.
“An experimental campaign supports our proposal, showing its effectiveness and viability.”
Quote in the paperSubstantial claimMay combine two claims.
Findingown data§ B. Uncovering novel Twitter spambots · p. 4In Test-set1, the LCS curve shows a continuous decrease and a sudden drop just before reaching 1,000 accounts; in Test-set2, there is a steep fall just before 400 accounts.
“Observing Figure 2(a), we can see that there is a continuous decrease in the LCS length as the number of accounts grows, and a sudden drop occurs just before reaching the 1,000 accounts. The drop is even more evident in Figure 2(b) where there is a steep fall of the curve just before 400 accounts.”
Quote in the paperSubstantial claimMay combine two claims.
Findingown data§ A. Characterizing spambot behavior · p. 3For both spambot groups, there is a sudden drop in LCS length when k gets close to the group size, whereas the human group has an exponential decay.
“For both the spambot groups we observe a sudden drop in LCS length when k gets close to the group size. Instead, the human group has an exponential decay and it rapidly reaches the smallest values of the LCS length.”
Quote in the paperSubstantial claimMay combine two claims.
Findingown data§ A. Characterizing spambot behavior · p. 3Tweet content DNA yields much lower LCS values than tweet type DNA; tweet type DNA models almost perfectly the high similarity of Bot2 accounts, while tweet content DNA cannot show a significant difference between genuine and bot accounts.
“All the LCS curves have much lower values, however the biggest difference between tweet content DNA and tweet type DNA is shown for Bot2 . While tweet type DNA models almost perfectly the high similarity of Bot2 accounts, tweet content DNA is unable to show a significative difference between genuine and bot accounts.”
Quote in the paperSubstantial claimMay combine two claims.
Premisecitation§ III. INTRODUCING DIGITAL DNA FINGERPRINTING · p. 2New waves of malicious accounts present different and advanced features, making their detection with existing systems extremely challenging.
“New waves of malicious accounts present different and advanced features, making their detection with existing systems extremely challenging [4], [6].”
Quote in the paperSubstantial claim0 of 2 sources checked- cites [4] [4] C. Yang, R. Harkreader, and G. Gu, 'Empirical evaluation and new design for fighting evolving twitter spammers,' Information Forensics and Security, IEEE Transactions on , vol. 8, no. 8, pp. 1280-1293, 2013. - source
Not checked in the free run.
- cites [6] [6] X. Hu, J. Tang, and H. Liu, 'Online social spammer detection,' in AAAI Conference on Artificial Intelligence , 2014, pp. 59-65. - source
Not checked in the free run.
May combine two claims.
Interpretationown data§ B. Uncovering novel Twitter spambots · p. 5This result supports the initial claim that the new wave of bots is surprisingly similar to genuine accounts and exceptionally hard to detect one by one.
“This result supports our initial claim that this new wave of bots is surprisingly similar to genuine accounts: they are exceptionally hard to detect if considered one by one.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).
Interpretationnone given§ V. CONCLUSION · p. 6The proposed methodology could be used in other contexts as well.
“Finally, we also argue that the proposed methodology could be used in other contexts as well, hence paving the way for further research.”
Quote in the paperNot a substantial claimMay combine two claims.
Interpretationreasoning§ B. Uncovering novel Twitter spambots · p. 3Properly extracted digital DNA sequences provide evidence of similar activities in large groups, which can lead to suspect a high degree of automation for accounts behaving similarly.
“As shown before, properly extracted digital DNA sequences provide evidence of similar activities in large groups. In turn, this can lead to suspect a high degree of automation for those accounts behaving in a similar way.”
Quote in the paperSubstantial claim- The authors state this as an assumption: Similarity in digital DNA is a proxy for automation.
Probably not a substantial claim: may not pass S2 (checkable).
Interpretationcitation§ B. Digital DNA as a proxy of user online behavior · p. 1Digital DNA is a flexible and compact way of modeling online user behaviors.
“Different kinds of user behaviors can be observed on the Internet [6] and digital DNA is a flexible and compact way of modeling such behaviors.”
- Also stated in: § B. Digital DNA as a proxy of user online behavior · p. 1 · plain
“Similarly to its biological predecessor, digital DNA is a compact representation of information.”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [6] [6] X. Hu, J. Tang, and H. Liu, 'Online social spammer detection,' in AAAI Conference on Artificial Intelligence , 2014, pp. 59-65. - source
Not checked in the free run.
- Also stated in: § B. Digital DNA as a proxy of user online behavior · p. 1 · plain
Probably not a substantial claim: may not pass S2 (checkable), S4 (based).
Premisenone given§ B. Digital DNA as a proxy of user online behavior · p. 1Many bioinformatics techniques developed for analyzing biological DNA can be carried over to study digital DNA.
“Indeed, many of the techniques developed in the field of bioinformatics for the analysis of biological DNA can be carried over to study the characteristics of digital DNA as well.”
Quote in the paperNot a substantial claimClassified as probably not a substantial claim.
Study design and conductnone given§ A. Characterizing spambot behavior · p. 3To be statistically sound, groups of the same size were compared by randomly undersampling genuine accounts to match the number of spambots.
“To be statistically sound, we compared groups of the same size by randomly undersampling genuine accounts to match the number of spambots.”
Quote in the paperSubstantial claimMay combine two claims.
Interpretationreasoning§ B. Uncovering novel Twitter spambots · p. 6The DNA fingerprinting technique only exploits Twitter timeline data, does not require a training phase, and can be employed in an unsupervised fashion.
“Instead, the DNA fingerprinting technique only exploits Twitter timeline data to perform spambots detection. Furthermore, it does not require a training phase and can be employed pretty much like a clustering algorithm, in an unsupervised fashion.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S4 (based).
Interpretationreasoning§ DNA-inspired online behavioral modeling and its application to spambot detection · p. 1The proposed methodology is platform and technology agnostic.
“While Twitter spambot detection is a specific use case on a specific social media, our proposed methodology is platform and technology agnostic, hence paving the way for diverse behavioral characterization tasks.”
Quote in the paperNot a substantial claimMay combine two claims.
Study design and conductown data§ A. Characterizing spambot behavior · p. 3All accounts in the datasets underwent manual verification; 991 Bot1 and 464 Bot2 accounts were certified as spambots, and 3,474 human accounts were certified as human.
“All the accounts of our datasets have undergone manual verification. Specifically, 50.05% (991 accounts) of Bot1 accounts and 89.29% (464 accounts) of Bot2 accounts were certified as spambots. Then, we built a dataset of human accounts by randomly contacting Twitter users and by asking them simple questions in natural language. All the 3,474 accounts that answered were certified as human.”
Quote in the paperSubstantial claim
The authors state this as an assumption.
Similarity in digital DNA is a proxy for automation.
“We consider similarity as a proxy for automation and, thus, an exceptionally high level of similarity among a large group of accounts serves as a red flag for anomalous behaviors.”
The authors state this as an assumption.
The suspiciously similar accounts are the ones that share the longest LCS just before the identified splitting points.
“we assumed that the suspiciously similar accounts are the ones that share the longest LCS just before such splitting points”
The paper relies on this without support.
The authors are the first to identify and adapt DNA-inspired techniques to online user behavioral modeling.
“To the best of our knowledge, we are the first ones to identify and adapt DNA-inspired techniques to online user behavioral modeling.”
The paper relies on this without support.
The definition of a unifying approach is an open challenge.
“The definition of a unifying approach is an open challenge.”
The authors state this as an assumption.
The basic assumption of digital DNA fingerprinting is to consider behaviors as sequences of actions and to characterize and detect social spambots by grouping similar sequences.
“The basic assumption of the digital DNA fingerprinting technique is to consider behaviors as sequences of actions and to characterize and detect social spambots by grouping similar sequences.”
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?
The digital DNA-based technique of Cresci et al. [14] achieves excellent detection performances.
“Results reported in Table 7 demonstrate that the digital DNA-based technique [14] achieves excellent detection performances.”
From The Paradigm-Shift of Social Spambots: Evidence, Theories, and Tools for the Arms Race (2017) · cites this paper as [14]
Although working at group level, the digital DNA technique of Cresci et al. [14] is capable of spotting single spambot accounts.
“Notably, although working at group level, [14] is capable of spotting single spambot accounts.”
From The Paradigm-Shift of Social Spambots: Evidence, Theories, and Tools for the Arms Race (2017) · cites this paper as [14]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Test-set1 detected group size differs from the results table
MinorThe text says 983 Test-set1 accounts were identified as suspicious, but the DNA fingerprinting row reports 963 true positives and 18 false positives, totaling 981 detected accounts.
“the 983 accounts with LCS around 400 for Test-set1”
“| DNA fingerprinting | 963 | 924 | 18 | 28 | 0.982”
Test-set2 detected group size differs from the results table
MinorThe text says 400 Test-set2 accounts were identified as suspicious, but the DNA fingerprinting row reports 398 true positives and no false positives, totaling 398 detected accounts.
“the 400 accounts with LCS of around 1750 for Test-set2”
“| DNA fingerprinting | 398 | 468 | 0 | 66 | 1.000”
Cites
Fame for sale: Efficient detection of fake Twitter followers
Cresci, S., Di Pietro, R., Petrocchi, M., Spognardi, A., & Tesconi, M. (2015) · Decision Support Systems · doi:10.1016/j.dss.2015.09.003 · Cited by 461 · see citing works
Self-citationEmpirical evaluation and new design for fighting evolving Twitter spammers
Yang, C., Harkreader, R., & Gu, G. (2013) · IEEE Transactions on Information Forensics and Security · doi:10.1109/tifs.2013.2267732 · Cited by 346 · cited 6× · see citing works
“[4] and the unsupervised approaches by Miller et al.”
BackgroundInfluentialTwitter spammer detection using data stream clustering
Miller, Z., Dickinson, B., Deitrick, W., Hu, W., & Wang, A. H. (2014) · Information Sciences · doi:10.1016/j.ins.2013.11.016 · Cited by 332 · cited 6× · see citing works
“To thoroughly evaluate the DNA fingerprinting technique we compared our detection results with those obtained by several different state-of-the-art approaches, namely the supervised one by Yang et al. [4] and the unsupervised approaches by Miller et al. [8] and by Ahmed et al. [9].”
BackgroundMethodResultInfluentialPrediction of Human Activity by Discovering Temporal Sequence Patterns
Li, K., & Fu, Y. (2014) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2013.2297321 · Cited by 212 · cited 3× · see citing works
“We envisage digital DNA sequences to embrace different applications, among which: • letting behavioral patterns emerge from the crowd [2], [5], by making use of standard DNA sequence alignment tools.”
BackgroundMethodA generic statistical approach for spam detection in Online Social Networks
Ahmed, F., & Abulaish, M. (2013) · Computer Communications · doi:10.1016/j.comcom.2013.04.004 · Cited by 199 · cited 8× · see citing works
“However, although achieving an overall good performance for the considered spambots, the approach in [9] might lack reusability across other groups of spambots with different behaviors, such as those perpetrating a follower fraud [5],[11].”
BackgroundMethodInfluentialOnline Social Spammer Detection
Hu, X., Tang, J., & Liu, H. (2014) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v28i1.8728 · Cited by 133 · cited 4× · see citing works
“Different kinds of user behaviors can be observed on the Internet [6] and digital DNA is a flexible and compact way of modeling such behaviors.”
BackgroundInfluentialCatching Synchronized Behaviors in Large Networks: A Graph Mining Approach
Jiang, M., Cui, P., Beutel, A., Faloutsos, C., & Yang, S. (2016) · ACM Transactions on Knowledge Discovery from Data · doi:10.1145/2746403 · Cited by 88 · cited 4× · see citing works
“Working with ordered and variable-length data representations also marks a difference with other techniques recently employed for the behavioral analysis of groups of users, such as those based on hashing [7] and graph mining [5].”
BackgroundMethodInfluentialPredicting Elections for Multiple Countries Using Twitter and Polls
Tsakalidis, A., Papadopoulos, S., Cristea, A. I., & Kompatsiaris, Y. (2015) · IEEE Intelligent Systems · doi:10.1109/mis.2015.17 · Cited by 79 · cited 2× · see citing works
“One is to mine substantial information regarding events of public interest [1].”
BackgroundLinear Time Algorithms for Generalizations of the Longest Common Substring Problem
Arnold, M., & Ohlebusch, E. (2009) · Algorithmica · doi:10.1007/s00453-009-9369-1 · Cited by 33 · cited 1× · see citing works
“We rely on a linear time algorithm [10] that is based on the generalized suffix tree.”
MethodUncovering deception in social media
Liu, H., Han, J., & Motoda, H. (2014) · Social Network Analysis and Mining · doi:10.1007/s13278-014-0162-z · Cited by 21 · cited 1× · see citing works
“Furthermore, online behavioral analysis helps in detecting fictitious and deceitful accounts that may distribute spam or lead to a bias in the public opinion [3].”
BackgroundProbabilistic Attributed Hashing
Ou, M., Cui, P., Wang, J., et al. (2015) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v29i1.9543 · Cited by 10 · cited 1× · see citing works
“Working with ordered and variable-length data representations also marks a difference with other techniques recently employed for the behavioral analysis of groups of users, such as those based on hashing [7] and graph mining [5].”
Method
Flagged first, then by citations
2 of the 13 references couldn’t be matched to a known paper.
Cited by
Sorted by citations
Tucker, J. A., Guess, A., Barbera, P., Vaccari, C., Siegel, A., Sanovich, S., Stukal, D., & Nyhan, B. (2018) · doi:10.2139/ssrn.3144139 · Cited by 1,222 · see citing works
Arming the public with artificial intelligence to counter social bots
Yang, K., Varol, O., Davis, C. A., Ferrara, E., Flammini, A., & Menczer, F. (2019) · Human Behavior and Emerging Technologies · doi:10.1002/hbe2.115 · Cited by 419 · see citing works
“Cresci et al. (2016) encoded tweet type or content as strings of symbols, then considered the longest common substring to detect coordination.”
BackgroundThe Paradigm-Shift of Social Spambots: Evidence, Theories, and Tools for the Arms Race
Cresci, S., Di Pietro, R., Petrocchi, M., Spognardi, A., & Tesconi, M. (2017) · Proceedings of the 26th International Conference on World Wide Web Companion - WWW '17 Companion · doi:10.1145/3041021.3055135 · Cited by 381 · see citing works
Self-citation(2020) · doi:10.1609/aaai.v34i01.5460 · Cited by 361 · see citing works
Detection of Bots in Social Media: A Systematic Review
Orabi, M., Mouheb, D., Aghbari, Z. A., & Kamel, I. (2020) · Information Processing & Management · doi:10.1016/j.ipm.2020.102250 · Cited by 232 · see citing works
The paradigm-shift of social spambots: Evidence, theories, and tools for the arms race
Cresci, S., Pietro, R. D., Petrocchi, M., Spognardi, A., & Tesconi, M. (2017) · Technical University of Denmark, DTU Orbit (Technical University of Denmark, DTU) · doi:10.48550/arxiv.1701.03017 · Cited by 237 · see citing works
Self-citationTwiBot-20: A Comprehensive Twitter Bot Detection Benchmark
Feng, S., Wan, H., Wang, N., Li, J., & Luo, M. (2021) · ACM International Conference on Information and Knowledge Management (CIKM) · doi:10.1145/3459637.3482019 · Cited by 199 · see citing works
“Other features are also adopted such as information on the user homepage [16], social networks [22], and timeline of accounts [5].”
Background(2020) · Sabanci University · Cited by 197 · see citing works
Social fingerprinting: detection of spambot groups through DNA-inspired behavioral modeling
Cresci, S., Di Pietro, R., Petrocchi, M., Spognardi, A., & Tesconi, M. (2017) · IEEE Transactions on Dependable and Secure Computing · doi:10.1109/tdsc.2017.2681672 · Cited by 168 · see citing works
Self-citationFeng, S., Wan, H., Wang, N., & Luo, M. (2021) · IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM) · doi:10.1145/3487351.3488336 · Cited by 172 · see citing works
“A large proportion of previous bot detection proposals [5], [7], [11], [34] have focused on leveraging these user property items and conduct effective feature engineering.”
BackgroundInfluentialTwitter Bot Detection Using Bidirectional Long Short-Term Memory Neural Networks and Word Embeddings
Wei, F., & Nguyen, U. T. (2019) · 2019 First IEEE International Conference on Trust, Privacy and Security in Intelligent Systems and Applications (TPS-ISA) · doi:10.1109/tps-isa48467.2019.00021 · Cited by 168 · see citing works
“Although our model performs slightly below the algorithm in [5] on some other metrics, our model offers many significant advantages over [5]: No handcrafted features required: Our model does not rely on any human-engineered features.”
BackgroundMethodResultInfluentialDetection of Novel Social Bots by Ensembles of Specialized Classifiers
Sayyadiharikandeh, M., Varol, O., Yang, K.-C., Flammini, A., & Menczer, F. (2020) · Proceedings of the 29th ACM International Conference on Information & Knowledge Management · doi:10.1145/3340531.3412698 · Cited by 159 · see citing works
“analysis of the activity of multiple accounts to reveal their coordination. Depending on the type of bots, similarity can be detected through tweet content [8, 25], temporal features in the timelines [7, 10], or retweeting behavior [28, 46]. A recent research direction is to address the limits of current bot detection…”
BackgroundDetecting Bots on Russian Political Twitter
Stukal, D., Sanovich, S., Bonneau, R., & Tucker, J. A. (2017) · Big Data · doi:10.1089/big.2017.0038 · Cited by 140 · see citing works
“Cresci et al. 30 propose to use a sequential data mining technique based on coding account activities over time with sequences of symbols, each representing a different type of activity, and finding the longest common substring within accounts’ strings.”
Uncovering Coordinated Networks on Social Media: Methods and Case Studies
Pacheco, D., Hui, P.-M., Torres-Lugo, C., Truong, B. T., Flammini, A., & Menczer, F. (2021) · Proceedings of the International AAAI Conference on Web and Social Media · doi:10.1609/icwsm.v15i1.18075 · Cited by 140 · see citing works
“A method named Digital DNA proposed to encode the tweet type or content as a string, which was then used to identify the longest common sub-string between accounts (Cresci et al. 2016).”
MethodDetect Me If You Can: Spam Bot Detection Using Inductive Representation Learning
Alhosseini, S. A., Tareaf, R. B., Najafi, P., & Meinel, C. (2019) · International World Wide Web Conference (WWW) · doi:10.1145/3308560.3316504 · Cited by 133 · see citing works
A one-class classification approach for bot detection on Twitter
Rodríguez‐Ruiz, J., Mata-Sánchez, J. I., Monroy, R., Loyola‐González, O., & López‐Cuevas, A. (2020) · Computers & Security · doi:10.1016/j.cose.2020.101715 · Cited by 121 · see citing works
Heterogeneity-Aware Twitter Bot Detection with Relational Graph Transformers
Feng, S., Tan, Z., Li, R., & Luo, M. (2022) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v36i4.20314 · Cited by 134 · see citing works
“Features extracted from tweets (Cresci et al. 2016) and user metadata (Yang et al.”
BackgroundMethodInfluentialMachine learning-based social media bot detection: a comprehensive literature review
Aljabri, M., Zagrouba, R., Shaahid, A., Alnasser, F., Saleh, A., & Alomari, D. M. (2023) · Social Network Analysis and Mining · doi:10.1007/s13278-022-01020-5 · Cited by 132 · see citing works
“Some analyses were able to detect spammers successfully using unsupervised learning methods for instance, (Cresci et al. 2016) put forth a novel behavioral-based unsupervised approach for spambots accounts detection, inspired by biological DNA.”
Cashtag Piggybacking: Uncovering Spam and Bot Activity in Stock Microblogs on Twitter
Cresci, S., Lillo, F., Regoli, D., Tardelli, S., & Tesconi, M. (2019) · ACM Transactions on the Web · doi:10.1145/3313184 · Cited by 105 · see citing works
“ems. 6.1 Digital DNA for social bot detection Inthissectionweemployastate-of-the-artbotandspamdetectionsystem, speci•callydevelopedfor spo−ing malicious group activities, to classify suspicious users [18, 20]. „e goal of this experiment is to assess whether users that shared/retweeted the suspicious peak tweets we pre…”
BackgroundMethodInfluentialSelf-citationUsing BERT to Extract Topic-Independent Sentiment Features for Social Media Bot Detection
Heidari, M., & Jones, J. H. (2020) · 2020 11th IEEE Annual Ubiquitous Computing, Electronics & Mobile Communication Conference (UEMCON) · doi:10.1109/uemcon51285.2020.9298158 · Cited by 102 · see citing works
“The last section of this work evaluates the proposed new bot detection model compared with existing state-of-the-art bot technology techniques in the same training and test data set of Cresci [27].”
MethodHeidari, M., Jones, J. H., & Uzuner, Ö. (2020) · IEEE ... International Conference on Data Mining workshops · doi:10.1109/icdmw51313.2020.00071 · Cited by 99 · see citing works
“TABLE VIII PERFORMANCE COMPARISON BETWEEN NEW BOT DETECTION MODEL AND PREVIOUS BOT DETECTION TECHNIQUES [23]”
Fake news outbreak 2021: Can we stop the viral spread?
Khan, T., Michalas, A., & Akhunzada, A. (2021) · Journal of Network and Computer Applications · doi:10.1016/j.jnca.2021.103112 · Cited by 91 · see citing works
Feng, S., Wan, H., Wang, N., Li, J., & Luo, M. (2021) · ACM International Conference on Information and Knowledge Management (CIKM) · doi:10.1145/3459637.3481949 · Cited by 96 · see citing works
“Other features are also adopted such as information on the user profile [25], social networks [31] and timeline of accounts [4].”
BackgroundContrast Pattern-Based Classification for Bot Detection on Twitter
Loyola‐González, O., Monroy, R., Rodríguez‐Ruiz, J., López‐Cuevas, A., & Mata-Sánchez, J. I. (2019) · IEEE Access · doi:10.1109/access.2019.2904220 · Cited by 80 · see citing works
“[39] proposed that not only are the actions but the sequence of the actions which allows dis-”
BackgroundMethodDeepSBD: A Deep Neural Network Model With Attention Mechanism for SocialBot Detection
Fazil, M., Sah, A. K., & Abulaish, M. (2021) · IEEE Transactions on Information Forensics and Security · doi:10.1109/tifs.2021.3102498 · Cited by 86 · see citing works
Pham, P., Nguyen, L. T. T., Vo, B., & Yun, U. (2021) · Information Systems · doi:10.1016/j.is.2021.101771 · Cited by 77 · see citing works
Unsupervised Social Bot Detection via Structural Information Theory
Peng, H. M., Zhang, J., Huang, X., Hao, Z., Li, A., Yu, Z., & Yu, P. S. (2024) · ACM Transactions on Information Systems · doi:10.1145/3660522 · Cited by 82 · see citing works
Zhao, C., Xin, Y., Li, X., Yang, Y., & Chen, Y. (2020) · Applied Sciences · doi:10.3390/app10030936 · Cited by 74 · see citing works
“[21] proposed a method similar to a clustering”
Method$FAKE: Evidence of Spam and Bot Activity in Stock Microblogs on Twitter
Cresci, S., Lillo, F., Regoli, D., Tardelli, S., & Tesconi, M. (2018) · Proceedings of the International AAAI Conference on Web and Social Media · doi:10.1609/icwsm.v12i1.15073 · Cited by 74 · see citing works
“In this section we employ a state-of-the-art bot and spam detection sys-tem – specifically developed for spotting malicious group activities – to classify suspicious users (Cresci et al. 2016; 2017b).”
MethodInfluentialSelf-citationA holistic system for troll detection on Twitter
Fornacciari, P., Mordonini, M., Poggi, A., Sani, L., & Tomaiuolo, M. (2018) · Computers in Human Behavior · doi:10.1016/j.chb.2018.08.008 · Cited by 69 · see citing works
A Large-scale Behavioural Analysis of Bots and Humans on Twitter
Gilani, Z., Farahbakhsh, R., Gareth Tyson, A., & Crowcroft, J. (2019) · ACM Transactions on the Web · doi:10.1145/3298789 · Cited by 62 · see citing works
“In another study, Cresci et al. used a similarity measure for social fingerprints to distinguish genuine users and spambots [19].”
BackgroundA Social Bots Detection Model Based on Deep Learning Algorithm
Ping, H., & Qin, S. (2018) · IEEE International Conference on Communication Technology (ICCT) · doi:10.1109/icct.2018.8600029 · Cited by 64 · see citing works
“While Twitter bot detection is a specific use case on a particular social media, Cresci's[10] approach is platformtechnology-independent online user behavior modeling: the digital DNA sequences are extracted and analyzed from the user's online actions, by comparing the digital DNA sequences to each other, and then the…”
BackgroundShahid, W., Jamshidi, B., Hakak, S., Isah, H., Khan, W. Z., Khan, M. K., & Choo, K. R. (2022) · IEEE Transactions on Computational Social Systems · doi:10.1109/tcss.2022.3177359 · Cited by 63 · see citing works
“For instance, Cresci et al. [76] proposed a simple and effective approach in which they extract digital DNA sequences from users’ online actions to model their behavior.”
BackgroundBayesian Graph Local Extrema Convolution with Long-tail Strategy for Misinformation Detection
Zhang, G., Zhang, S., & Yuan, G. (2024) · ACM Transactions on Knowledge Discovery from Data · doi:10.1145/3639408 · Cited by 57 · see citing works
“Accuracy F1-score MCC DSC [41] 0.4801 0.6266 -0.1372 Digital DNA [9] 0.4793 0.1072 0.0839 BotOrNot [10] 0.5584 0.4892 0.1558 Alhosseini et al. [2] 0.6813 0.7318 0.3543 SATAR [11] 0.8412 0.8642 0.6863 RGCN [13] 0 — SATAR: SATAR [11] is a self-supervised algorithm that uses the user’s semantic,…”
InfluentialBetter Safe Than Sorry: An Adversarial Approach to Improve Social Bot Detection
Cresci, S., Petrocchi, M., Spognardi, A., & Tognazzi, S. (2019) · Proceedings of the 10th ACM Conference on Web Science · doi:10.1145/3292522.3326030 · Cited by 52 · see citing works
“As such accounts put in place complex mechanisms to evade existing detection systems, scholars tried to keep pace by proposing powerful techniques based on profile- [5, 10, 54], posting- [5, 8, 11, 13, 22, 48], and networkcharacteristics [5, 31, 47, 48, 51–53] of the accounts.”
BackgroundMethodResultInfluentialSelf-citationOn the capability of evolved spambots to evade detection via genetic engineering
Cresci, S., Petrocchi, M., Spognardi, A., & Tognazzi, S. (2018) · Online Social Networks and Media · doi:10.1016/j.osnem.2018.10.005 · Cited by 53 · see citing works
Self-citationSimilCatch: Enhanced social spammers detection on Twitter using Markov Random Fields
El-Mawass, N., Honeiné, P., & Vercouter, L. (2020) · Information Processing & Management · doi:10.1016/j.ipm.2020.102317 · Cited by 53 · see citing works
“This process is similar to the one described in Cresci et al. (2016); El-Mawass515 13Compromise is fairly common on social media.”
ResultPolitical Bots and the Swedish General Election
Fernquist, J., Kaati, L., & Schroeder, R. (2018) · IEEE International Conference on Intelligence and Security Informatics (ISI) · doi:10.1109/isi.2018.8587347 · Cited by 57 · see citing works
“In Table II we notice that the model from [5] performs best of all the metrics except for recall where our model performs better.”
MethodEmergent properties, models, and laws of behavioral similarities within groups of twitter users
Cresci, S., Di Pietro, R., Petrocchi, M., Spognardi, A., & Tesconi, M. (2020) · Computer Communications · doi:10.1016/j.comcom.2019.10.019 · Cited by 40 · see citing works
Self-citationAvvenuti, M., Bellomo, S., Cresci, S., Polla, M. N. L., & Tesconi, M. (2017) · International World Wide Web Conference (WWW) · doi:10.1145/3041021.3051155 · Cited by 42 · see citing works
“In addition to the baselines derived from recent research in Q&A, we obtained a new baseline by measuring the users’ reaction to messages sent by a set of advanced Twitter spambots, thoroughly studied and described in [8].”
BackgroundMethodSelf-citationSystematic Literature Review of Social Media Bots Detection Systems
Ellaky, Z., Benabbou, F., & Ouahabi, S. (2023) · Journal of King Saud University - Computer and Information Sciences · doi:10.1016/j.jksuci.2023.04.004 · Cited by 38 · see citing works
“To deal with the fact that bots can evade detection by random-izing their actions, the authors of this work (Cresci et al., 2016) proposed a DNA-inspired model to detect spam bots.”
InfluentialAn Attention-Based Graph Neural Network for Spam Bot Detection in Social Networks
Zhao, C., Xin, Y., Li, X., Zhu, H., Yang, Y., & Chen, Y. (2020) · Applied Sciences · doi:10.3390/app10228160 · Cited by 36 · see citing works
“In the study of Cresci et al. [11], inspired by DNA sequences, according to the type of tweets, each account is modeled as a series of behavior data, and the corresponding “digital DNA” signature is generated.”
BackgroundBIC: Twitter Bot Detection with Text-Graph Interaction and Semantic Consistency
Lei, Z., Wan, H., Zhang, W., Feng, S., Chen, Z., Li, J., Zheng, Q., & Luo, M. (2023) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2023.acl-long.575 · Cited by 45 · see citing works
Bots, disinformation, and the first impeachment of U.S. President Donald Trump
Rossetti, M., & Zaman, T. R. (2023) · PLoS ONE · doi:10.1371/journal.pone.0283971 · Cited by 34 · see citing works
Detecting inorganic financial campaigns on Twitter
Tardelli, S., Avvenuti, M., Tesconi, M., & Cresci, S. (2021) · Information Systems · doi:10.1016/j.is.2021.101769 · Cited by 31 · see citing works
Self-citationReady-to-(ab)use: From fake account trafficking to coordinated inauthentic behavior on Twitter
Mazza, M., Cola, G., & Tesconi, M. (2022) · Online Social Networks and Media · doi:10.1016/j.osnem.2022.100224 · Cited by 32 · see citing works
Self-citationHays, C., Schutzman, Z., Raghavan, M., Walk, E., & Zimmer, P. (2023) · International World Wide Web Conference (WWW) · doi:10.1145/3543507.3583214 · Cited by 29 · see citing works
“Finally, a fourth approach seeks insight from other disciplines by using behavioral [32, 36] or biology-inspired techniques [15–17 , 60 Limitations of bot detection tools.”
BackgroundCresci, S., Petrocchi, M., Spognardi, A., & Tognazzi, S. (2018) · International World Wide Web Conference (WWW) · doi:10.1145/3184558.3191595 · Cited by 28 · see citing works
“Our ambitious vision is motivated by the combination of the following three proven facts: (i) Recently, a behavioral modeling technique inspired by biological DNA has been proposed – and successfully applied – for monitoring and detecting spambots in OSNs [2, 4].”
BackgroundSelf-citationUncovering Coordinated Networks on Social Media.
Pacheco, D. F., Hui, P.-M., Torres-Lugo, C., Truong, B. T., Flammini, A., & Menczer, F. (2020) · Cited by 29 · see citing works
Li, S., Zhao, C., Li, Q., Huang, J., Zhao, D., & Zhu, P. (2022) · World Wide Web · doi:10.1007/s11280-022-01114-2 · Cited by 31 · see citing works
“[11-12] proposed a revised approach based on DNA-inspired techniques in order to model online user behavior.”
MethodTwitter User Profiling: Bot and Gender Identification
Kosmajac, D., & Kešelj, V. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-58219-7_13 · Cited by 26 · see citing works
On the need of opening up crowdsourced emergency management systems
Avvenuti, M., Cresci, S., Vigna, F. D., et al. (2017) · AI & Society · doi:10.1007/s00146-017-0709-4 · Cited by 25 · see citing works
“…several applications feeding on SM data have been developed with the purpose of event detection, disaster response (Avvenuti et al. 2016c), business intelligence, health monitoring (Del Vigna et al. 2016a), cybercrime fight (Cresci et al. 2016), and situational awareness (Cresci et al. 2015a).”
BackgroundSelf-citationOn the efficacy of old features for the detection of new bots
Nicola, R. D., Petrocchi, M., & Pratelli, M. (2021) · Information Processing & Management · doi:10.1016/j.ipm.2021.102685 · Cited by 24 · see citing works
“Approximately in the last five years, say up to 2019, instead, a number of research teams independently have proposed new approaches that aim at detecting coordinated and synchronized behavior of groups of automated malicious accounts, see, e.g., [16, 17, 18].”
Self-citationLuceri, L., Pantè, V., Burghardt, K., & Ferrara, E. (2024) · International World Wide Web Conference (WWW) · doi:10.1145/3589334.3645529 · Cited by 25 · see citing works
“For bots, solutions have used various features and machine learning strategies to identify bot characteristics [8, 10, 57].”
MethodTwitter Bot Detection with Reduced Feature Set
Abreu, J. V. F., Ralha, C. G., & Gondim, J. J. C. (2020) · IEEE International Conference on Intelligence and Security Informatics (ISI) · doi:10.1109/isi49825.2020.9280525 · Cited by 23 · see citing works
“The results of the classifier were compared with five other works: [21], [24], [25], [26], [27] achieving recall of 0.976, which is greater than all five compared works (the second-largest was [22] with 0.972).”
ResultInfluentialHHG-Bot: A Hyperheterogeneous Graph-Based Twitter Bot Detection Model
Wang, T., Wang, Z., Li, H., Xia, C., & Zhao, C. J. (2025) · IEEE Transactions on Computational Social Systems · doi:10.1109/tcss.2025.3543419 · Cited by 27 · see citing works
(2017) · doi:10.1109/dsaa.2017.57 · Cited by 21 · see citing works
“heterogeneity in legitimate behaviors [5], [7].”
BackgroundResultInfluentialSelf-citationBeyond Text: Multimodal Credibility Assessment Approaches for Online User-Generated Content
Choudhary, M., Chouhan, S. S., & Rathore, S. S. (2024) · ACM Transactions on Intelligent Systems and Technology · doi:10.1145/3673236 · Cited by 20 · see citing works
Detection of Fake Twitter Accounts with Machine Learning Algorithms
Aydın, İ., Sevi, M., & Salur, M. U. (2018) · International Artificial Intelligence and Data Processing Symposium (IDAP) · doi:10.1109/idap.2018.8620830 · Cited by 20 · see citing works
“Cyborg hesaplarının büyük bir kısmı ticari şirketler ve web siteleri tarafından yeni bir medya kanalı ve müşteri hizmeti türü olarak kayıtlıdırlar [7].”
Should we agree to disagree about Twitter’s bot problem?
Varol, O. (2023) · Online Social Networks and Media · doi:10.1016/j.osnem.2023.100263 · Cited by 20 · see citing works
Identifying active, reactive, and inactive targets of socialbots in Twitter
Fazil, M., & Abulaish, M. (2017) · IEEE/WIC/ACM International Conference on Web Intelligence · doi:10.1145/3106426.3106483 · Cited by 17 · see citing works
“Another sequence-based approach inspired from DNA sequencing is used in [12] for profiling individual activities that are grouped-by activity-type and activity-sequence to segregate social spambots from benign users.”
BackgroundTwitter Bot Detection Using Bidirectional Long Short-term Memory Neural Networks and Word EmbeddingsPreprint
Feng, W. Q., & Nguyen, U. T. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2002.01336 · Cited by 18 · see citing works
“Although our model performs slightly below the algorithm in [5] on some other metrics, our model offers many significant advantages over [5]: No handcrafted features required: Our model does not rely on any human-engineered features.”
BackgroundMethodResultInfluentialExposing Cross-Platform Coordinated Inauthentic Activity in the Run-Up to the 2024 U.S. Election
Cinus, F., Minici, M., Luceri, L., et al. (2025) · doi:10.1145/3696410.3714698 · Cited by 17 · see citing works
“10 Both RT.com and Ruptly.tv have previously been accused of orchestrating campaigns to influence U.S. Elections via social media 11 .”
BotTriNet: A Unified and Efficient Embedding for Social Bots Detection via Metric Learning
Wu, J., Ye, X., & Man, Y. (2023) · International Symposium on Digital Forensics and Security (ISDFS) · doi:10.1109/isdfs58141.2023.10131839 · Cited by 17 · see citing works
“36% on the real-world dataset CRESCI2017, surpassing the performance of prior works such as [14], [15], [16], and [8].”
BackgroundAn Evolutionary Computation Approach for Twitter Bot Detection
Rovito, L., Bonin, L., Manzoni, L., & Lorenzo, A. D. (2022) · Applied Sciences · doi:10.3390/app12125915 · Cited by 18 · see citing works
Sybil Detection in Social-Activity Networks: Modeling, Algorithms and Evaluations
Zhang, X., Xie, H. L., & Lui, J. C. S. (2018) · IEEE International Conference on Network Protocols (ICNP) · doi:10.1109/icnp.2018.00015 · Cited by 17 · see citing works
“Unsupervised detection usually clusters sybils according to different features, such as loosely synchronized actions [29], activity patterns [30], users’ profiles and tweets [31], click-streams [32], etc.”
BackgroundMultimodal Detection of Bots on X (Twitter) Using Transformers
Ilias, L., Kazelidis, I. M., & Askounis, D. (2024) · IEEE Transactions on Information Forensics and Security · doi:10.1109/tifs.2024.3435138 · Cited by 17 · see citing works
“This technique, which addresses the issue of imbalanced dataset by downsampling the set of real users, has been adopted by previous studies [15], [37].”
BackgroundMethodINFOSHIELD: Generalizable Information-Theoretic Human-Trafficking Detection
Lee, M.-C., Vajiac, C., Kulshrestha, A., Lévy, S., Park, N., Jones, C., Reihaneh Rabbany, K., & Faloutsos, C. (2021) · IEEE International Conference on Data Engineering (ICDE) · doi:10.1109/icde51399.2021.00101 · Cited by 17 · see citing works
“On Twitter data, we compare to three supervised methods [45], [14], [46] and one unsupervised method [47].”
MethodDomain-Aware Federated Social Bot Detection with Multi-Relational Graph Neural Networks
Peng, H., Zhang, Y., Sun, H., Bai, X., Li, Y., & Wang, S. (2022) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn55064.2022.9892366 · Cited by 18 · see citing works
“Early bot detection methods extract features from user profiles [3]–[5], tweets [6], etc. and apply the traditional classifiers, such as Random Forest models [7], to detect bots in social networks.”
BackgroundMethodCashtag piggybacking: uncovering spam and bot activity in stock microblogs on TwitterPreprint
Cresci, S., Lillo, F., Regoli, D., Tardelli, S., & Tesconi, M. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1804.04406 · Cited by 18 · see citing works
“ems. 6.1 Digital DNA for social bot detection Inthissectionweemployastate-of-the-artbotandspamdetectionsystem, speci•callydevelopedfor spo−ing malicious group activities, to classify suspicious users [18, 20]. „e goal of this experiment is to assess whether users that shared/retweeted the suspicious peak tweets we pre…”
BackgroundMethodInfluentialSelf-citationEnhancing Sybil Detection via Social-Activity Networks: A Random Walk Approach
Zhang, X., Xie, H. L., Yi, P., & Lui, J. C. S. (2022) · IEEE Transactions on Dependable and Secure Computing · doi:10.1109/tdsc.2022.3151701 · Cited by 15 · see citing works
“Unsupervised detection usually clusters sybils according to different features, such as loosely synchronized actions [30], activity patterns [31], users’ profiles and tweets [32], click-streams[33], etc.”
BackgroundTwitter Bot Detection Using Neural Networks and Linguistic Embeddings
Feng, W. Q., & Nguyen, U. T. (2023) · IEEE Open Journal of the Computer Society · doi:10.1109/ojcs.2023.3302286 · Cited by 17 · see citing works
“[5] in some cases, it offers many significant advantages over [5] as follows.”
BackgroundMethodInfluentialMulti-modal Social Bot Detection: Learning Homophilic and Heterophilic Connections Adaptively
Li, S., Qiao, B., Li, K., Lu, Q., Lin, M., & Zhou, W. J. (2023) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3581783.3612569 · Cited by 19 · see citing works
“Early social bot detection methods extracted user semantic and property information from tweets and meta-data through feature engineering for detection[9].”
MethodFake Profile Detection Using Machine Learning
Umbrani, K., Shah, D. S., Pile, A., et al. (2024) · ASU International Conference in Emerging Technologies for Sustainability and Intelligent Systems (ICETSIS) · doi:10.1109/icetsis61505.2024.10459570 · Cited by 15 · see citing works
“After cleaning all the data, we used TF-IDF vectorizer [12] to convert the categorical data into numerical data.”
MethodAkhtar, M. M., Masood, R., Ikram, M., & Kanhere, S. S. (2024) · ACM Asia Conference on Computer and Communications Security (ASIA CCS) · doi:10.1145/3634737.3644998 · Cited by 16 · see citing works
“Similarly, Cresci et al.[24, 28] proposed an innovative method that uses digital DNA sequences of the user’s timeline data to detect coordinating bots.”
MethodAre We Secure from Bots? Investigating Vulnerabilities of Botometer
Torusdağ, M. B., Kutlu, M., & Selçuk, A. A. (2020) · doi:10.1109/ubmk50275.2020.9219433 · Cited by 15 · see citing works
It’s a Matter of Style: Detecting Social Bots through Writing Style Consistency
Cardaioli, M., Conti, M., Sorbo, A. D., Fabrizio, E., Laudanna, S., & Visaggio, C. A. (2021) · International Conference on Computer Communications and Networks (ICCCN) · doi:10.1109/icccn52240.2021.9522339 · Cited by 15 · see citing works
Distance-based customer detection in fake follower markets
Jang, B., Jeong, S., & Kim, C.-K. (2018) · Information Systems · doi:10.1016/j.is.2018.12.001 · Cited by 14 · see citing works
Better Safe Than Sorry: An Adversarial Approach to Improve Social Bot DetectionPreprint
Cresci, S., Petrocchi, M., Spognardi, A., & Tognazzi, S. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1904.05132 · Cited by 17 · see citing works
“As such accounts put in place complex mechanisms to evade existing detection systems, scholars tried to keep pace by proposing powerful techniques based on profile- [5, 10, 54], posting- [5, 8, 11, 13, 22, 48], and networkcharacteristics [5, 31, 47, 48, 51–53] of the accounts.”
BackgroundMethodResultInfluentialSelf-citationPEGNN: Peripheral-Enhanced graph neural network for social bot detection
Guyan, Q., Liu, Y., Liu, J., et al. (2025) · Expert Systems with Applications · doi:10.1016/j.eswa.2025.127294 · Cited by 13 · see citing works
Characterizing Social Spambots by their Human Traits
Giorgi, S., Ungar, L., & Schwartz, H. A. (2021) · Findings of the Association for Computational Linguistics: ACL (Findings of ACL) · doi:10.18653/v1/2021.findings-acl.457 · Cited by 13 · see citing works
“Cresci et al.: uses DNA-inspired techniques to model behavior of Twitter users in an unsupervised fashion (Cresci et al., 2016).”
BackgroundMethodInfluentialBlackmarket-Driven Collusion on Online Media: A Survey
Dutta, H. S., & Chakraborty, T. (2021) · ACM/IMS Transactions on Data Science · doi:10.1145/3517931 · Cited by 13 · see citing works
“[35] showed how group of spambots exhibit common patterns as compared to group of genuine accounts.”
BackgroundTowards the future of bot detection: A comprehensive taxonomical review and challenges on Twitter/X
Javed, D., Jhanjhi, N. Z., Khan, N. A., Ray, S. K., Mazroa, A. A., Ashfaq, F., & Das, S. R. (2024) · Computer Networks · doi:10.1016/j.comnet.2024.110808 · Cited by 13 · see citing works
MRLBot: Multi-Dimensional Representation Learning for Social Media Bot Detection
Zeng, F., Sun, Y., & Li, Y. (2023) · Electronics · doi:10.3390/electronics12102298 · Cited by 12 · see citing works
“Cresci et al. [19,20] extracted and analyzed digital DNA sequences from users’ online behaviors, and applied standard DNA analysis techniques to distinguish between legitimate accounts and spam accounts on the platform. for unsupervised pretraining on a large number of user samples to Mazza et al.…”
MethodSupervised Classification of Social Spammers using a Similarity-based Markov Random Field Approach
El-Mawass, N., Honeiné, P., & Vercouter, L. (2018) · Multidisciplinary International Social Networks Conference (MISNC) · doi:10.1145/3227696.3227712 · Cited by 12 · see citing works
“This process is similar to what Cresci et al. describe in their paper [10].”
ResultTwitter Bot Detection using Diversity Measures
Kosmajac, D., & Kešelj, V. (2019) · Cited by 15 · see citing works
Socialbots: Impacts, Threat-Dimensions, and Defense Challenges
Abulaish, M., & Fazil, M. (2020) · IEEE Technology and Society Magazine · doi:10.1109/mts.2020.3012327 · Cited by 11 · see citing works
“However, as approaches become mature and advanced, spammers [21], [23].”
BackgroundMethodInfluentialGiglietto, F., Marino, G., Mincigrucci, R., & Stanziano, A. (2023) · Social Media + Society · doi:10.1177/20563051231196866 · Cited by 11 · see citing works
“Computational meth-ods are used to detect social bots by examining account information or their network of ties (Cresci et al., 2016; Davis et al., 2016; Wu et al., 2020).”
Chang, Y.-Y., Wang, W., & Peng, W.-C. (2024) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v38i1.27752 · Cited by 11 · see citing works
Javed, D., Jhanjhi, N. Z., Khan, N. A., Ray, S. K., Al-Dhaqm, A., & Kebande, V. R. (2025) · IEEE Access · doi:10.1109/access.2025.3551993 · Cited by 13 · see citing works
Misinformation Detection and Adversarial Attack Cost Analysis in Directional Social Networks
Shao, H., Yao, S., Jing, A., Liu, S., Liu, D., Wang, T., Li, J., Yang, C., Wang, R., & Abdelzaher, T. (2020) · International Conference on Computer Communications and Networks (ICCCN) · doi:10.1109/icccn49398.2020.9209609 · Cited by 11 · see citing works
“• DNA : This algorithm develops the digital DNA sequences to model online user behavior with a string [24].”
MethodDispelling the Fake: Social Bot Detection Based on Edge Confidence Evaluation
Qiao, B., Zhou, W. J., Li, K., Li, S., & Hu, S. (2024) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2024.3396192 · Cited by 12 · see citing works
“For posting activities, statistical analysis [36], [37], [38] and deoxyribonucleic acid (DNA) biosequence [39], [40], [41] methods have been commonly employed.”
An Enhanced Sybil Guard to Detect Bots in Online Social Networks
Shetty, N. P., Muniyal, B., Anand, A., et al. (2021) · Journal of Cyber Security and Mobility · doi:10.13052/jcsm2245-1439.1115 · Cited by 10 · see citing works
“Cresci et al. (2016) [21] incorporated DNA Analysis techniques to identify spam bots by building a profile for genuine users and identifying the subsequent variations.”
BackgroundLost in Transformation: Rediscovering LLM-Generated Campaigns in Social Media
Britta, G., Pohl, J., Winkelmann, H., Stampe, L., & Grimme, C. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-47896-3_6 · Cited by 9 · see citing works
Opinion Dynamic Modeling of News Perception
Milli, L. (2021) · Applied Network Science · doi:10.1007/s41109-021-00412-4 · Cited by 9 · see citing works
A Survey of Spam Bots Detection in Online Social Networks
Ellaky, Z., Benabbou, F., Ouahabi, S., & Sael, N. (2021) · International Conference on Digital Age & Technological Advances for Sustainable Development (ICDATA) · doi:10.1109/icdata52997.2021.00021 · Cited by 11 · see citing works
DNA-influenced automated behavior detection on twitter through relative entropy
Gilmary, R., Akila, V., Vaiyapuri, G., & Balamurali, D. (2022) · Scientific Reports · doi:10.1038/s41598-022-11854-w · Cited by 9 · see citing works
“Our technique is compared with the entropy-based approach on temporal patterns 19 , the DNA modeling-based research 13,14,16 , and the bot detection tool, Botometer 42,43 .”
BackgroundMethodInfluentialA hybrid framework for bot detection on twitter: Fusing digital DNA with BERT
Chawla, V., & Kapoor, Y. (2023) · Multimedia Tools and Applications · doi:10.1007/s11042-023-14730-5 · Cited by 10 · see citing works
“The optimal classification threshold works by detecting the slope of the LCS curve that represents the steepest drop [18 – 21] of the curve.”
BackgroundMethodResultInfluentialAlsodi, O., Zhou, X., Gururajan, R., Shrestha, A., & Btoush, E. (2025) · Applied Sciences · doi:10.3390/app15073898 · Cited by 10 · see citing works
Malicious accounts detection from online social networks: a systematic review of literature
Sassi, I. B., & Yahia, S. B. (2021) · International Journal of General Systems · doi:10.1080/03081079.2021.1976773 · Cited by 8 · see citing works
“…Boreggah et al. (2018), Cai et al. (2019), Chavoshi andMueen (2018), Chavoshi, Hamooni, andMueen (2016), Chu et al. (2012), Concone et al. (2019), Cresci et al. (2016), Cresci et al. (2018), David, Siordia, and Moctezuma (2016), Dickerson, Kagan, and Subrahmanian (2014), Dorri, Abadi, and…”
BackgroundInfluential
- Authors
- Cresci, S., Di Pietro, R., Petrocchi, M., Spognardi, A., & Tesconi, M.
- Affiliations
- Institute of Informatics and Telematics (IT)
- Technical University of Denmark (DK)
- Venue
- IEEE Intelligent Systems
- Published
- 2016-09
- Volume
- 31
- Issue
- 5
- Pages
- 58-64
- ISSN
- 1541-1672 (ISSN-L)
- 1941-1294
- OpenAlex
- W2269924745
- Semantic Scholar
- 44ae3eddbf938e1d500edfbf9354e0fa9699860a
- Cited by
- 194
- References
- 13
- Citations by year
- 2026: 14 · 2025: 15 · 2024: 20 · 2023: 32 · 2022: 19 · 2021: 21 · 2020: 34 · 2019: 18 · 2018: 12 · 2017: 8 · 2016: 1
- FWCI
- 29.09
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- IEEE Intelligent Systems (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- CINECA IRIS Institutial research information system (University of Pisa) (submittedVersion)
- IRIS Research product catalog (Sapienza University of Rome) (submittedVersion)
- Technical University of Denmark, DTU Orbit (Technical University of Denmark, DTU) (submittedVersion, other-oa)
- Padua Research Archive (University of Padova) (submittedVersion)
- arXiv (Cornell University)
- Topics
- Physical Sciences › Computer Science › Information Systems › Spam and Phishing Detection
- Physical Sciences › Computer Science › Information Systems › User Authentication and Security Systems
- Physical Sciences › Computer Science › Artificial Intelligence › Hate Speech and Cyberbullying Detection
- Keywords
- Twitter, social bots, behavioral characterization
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.
Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-09.
Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-09.
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
