Skip to main content

The tinder swindler: Analyzing public sentiments of romance fraud using machine learning and artificial intelligence

Lokanan, ME. (2023) — Journal of Economic Criminology

DOI: 10.1016/j.jeconc.2023.100023
Type:   Journal Article
Country:   Canada
Tags:   romance fraud, Tinder Swindler, Twitter, sentiment analysis, machine learning, BERT, victim blaming, impression management

Synopsis (AI-Generated)

This study analyzed 47,876 Twitter posts from 29,630 unique users containing #tinderswindler between February 1 and March 18, 2022. It used TextBlob to classify polarity and subjectivity and compared several machine-learning and neural sentiment models, framing interpretation through Goffman's impression-management theory. The author reports that classifications were predominantly neutral or positive and relatively infrequently negative; negative vocabulary also included victim-blaming terms portraying the defrauded women as desperate, unintelligent, or deserving of harm. These labels describe discourse around one documentary and hashtag, not public approval of romance fraud generally. Sarcasm, missing conversational context, algorithmic labeling error, and self-selected Twitter participation limit interpretation of individual posts and broader public sentiment.

Identified Gaps (AI-Generated)

The paper identifies a gap in applying Goffman’s impression-management theory to relational dynamics in romance fraud and states that no prior work had analyzed fraudsters’ actions through a Goffmanesque narrative lens. It also notes a need for more rigorous scientific analysis of self-presentation on social media. Methodologically, sentiment analysis has difficulty interpreting contextual meaning, irony, and sarcasm.

Methods (AI-Generated)

The study collected 47,876 Twitter posts using #tinderswindler from 1 February to 18 March 2022 (29,630 unique users). Tweets were cleaned, anonymized, tokenized, stemmed/lemmatized, and vectorized with bag-of-words and TF-IDF. TextBlob generated polarity and subjectivity classifications. Sentiment prediction compared Naïve Bayes, SVM, random forest, CNN, RNN+LSTM, and BERT models using a 60:20:20 train-validation-test split. The analysis was framed by Goffman’s impression-management theory.

Limitations (AI-Generated)

Sentiment is subjective and a tweet may reflect personal experience, irrationality, or sarcasm. Although the large aggregate sample was used to dilute outliers and bias, sentiment analysis can analyze text without adequate context, and algorithms struggle to distinguish ironic or sarcastic content. These issues caution against interpreting individual tweet-level classifications as definitive evidence of users’ views.

Future Work (AI-Generated)

Develop NLP/ML methods that can separate irrational and sarcastic tweets and preserve relevant textual context. Refine and optimize deep-learning approaches for sentiment analysis of romance-fraud discussions. Apply sentiment findings to test targeted awareness, moderation, and dating-platform interventions addressing manipulation and victim blaming.

Related Research Insights

See how this publication connects to RSRC's living evidence syntheses through current citations and research-topic mapping.

AI-Generated Content Notice

The synopsis and research notes on this page were generated with AI from available publication information and, when available, the uploaded paper text. They may contain errors, omissions, or interpretation issues. Readers should follow the DOI or source link, review the original publication, and make their own judgment about the content.

Found a possible error? Request a correction.