A Metric to Assist in Detecting International Phishing or Ransomware Cyberattacks
Bibliographic record
Abstract
Over the past decade, the number of cyberattacks such as ransomware, phishing, and other forms of malware have increased significantly, as has the danger to innocent users. The ability to launch such devastating attacks is no longer limited to well-funded, highly structured organizations including government agencies whose missions may well include cyberattacks.The focus of our study is threats to an individual not from such highly organized institutions, but rather less organized cybercriminal organizations with limited resources.The Internet provides ample opportunities for such criminal organizations to launch cyberattacks at minimal cost. One tool for such lower-level criminal organizations is Google Translate (GT) needed to launch a cyberattack on a user in a relatively advantaged country such as the United States, United Kingdom, or Canada. It has been observed that many such attacks may originate in a lesser developed country (LDC), where the local language is a language not common persons in target countries, for example English.It is a reasonable assumption that informal cyberattackers may not have a command of English and to use English for an attack online they may require a mechanism, such as the no-cost GT.In previous work, a number of authors have attempted to develop an index to measure the efficiency or what might be called an ABA translation. This involves beginning with a test document in language, then GT to translate into language then back again to A. The resulting original text is then compared to the transformation by using a modified Levenshtein distance computation for the A versions.The paper analyzes the process of determining an index to detect if a text has been translated from an original language and location, assuming the attack document has been written in one language and translated using GT into the language of the person attacked. The steps involved in this analysis include:a) Consistency: in order to determine consistency in the use of the ABA/GT process, the primary selection of test is compared with random samples from the test media;b) Expanded selection of languages for translation: prior work has established use of the technique for 12 language pairs. The current work extends analysis to a wider set of languages, including those reported as having the highest levels of cyberattacks.c) Back translation of selected languages: used to extend the quality of those translations are made.d) New language pairs are considered: by analyzing the countries and indigenous languages of the countries paired with the highest levels of cyberattack and the highest levels of cyberdefense, additional language pairs are added to this analysis;e) Comparison to prior results: results found in this paper are used for a proposed network for all language pairs considered in this analysis.The end product is a metric giving a probability of determining the original source language of the cyberattack as compared to the translation to the victim's language, with the expectation that this will allow for an increased likelihood of being able to identify the attackers.
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 | 0.021 |
| Meta-epidemiology (narrow) | 0.002 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.016 | 0.010 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.003 | 0.005 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.002 | 0.001 |
| Insufficient payload (model declined to judge) | 0.005 | 0.006 |
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".