Towards a Linguistic Stylometric Model for the Authorship Detection in Cybercrime Investigations
Bibliographic record
Abstract
This study proposes an integrated framework that considers letter-pair frequencies/combinations along with the lexical features of documents as a means to identifying the authorship of short texts posted anonymously on social media. Taking a quantitative morpho-lexical approach, this study tests the hypothesis that letter information, or mapping, can identify unique stylistic features. As such, stable word combinations and morphological patterns can be used successfully for authorship detection in relation to very short texts. This method offers significant potential in the fight against online hate speech, which is often posted anonymously and where authorship is difficult to identify. The data analyzed is from a corpus of 12,240 tweets derived from 87 Twitter accounts. A self-organizing map (SOM) model was used to classify input patterns in the tweets that shared common features. Tweets grouped in a particular class displayed features that suggested they were written by a particular author. The results indicate that the accuracy of classification according to the proposed system was around 76%. Up to 22% of this accuracy was lost, however, when only distinctive words were used and 26% was lost when the classification procedure was based solely on letter combinations and morphological patterns. The integration of letter-pairs and morphological patterns had the advantage of improving accuracy when determining the author of a given tweet. This indicates that the integration of different linguistic variables into an integrated system leads to better performance in classifying very short texts. It is also clear that the use of a self-organizing map (SOM) led to better clustering performance because of its capacity to integrate two different linguistic levels for each author profile.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.004 | 0.016 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.009 | 0.004 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.004 | 0.004 |
| Open science | 0.001 | 0.002 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".