Notice: This website is continuously updated as new conference proceedings and journal volumes are published. Consequently, the figures presented here may differ from those reported in our publications.
A brief overview of our methodology is provided on the Methods page, along with references to publications documenting our approach and links to the code and data used in each study.
The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates.
Optimal Personalized Filtering Against Spear-Phishing Attacks
Authors: Aron Laszka, Yevgeniy Vorobeychik, Xenofon Koutsoukos
AAAI 2015 | Venue PDF | LLM Run Details | Input Tokens: 14,788 Total number of tokens sent to the LLM as input for this paper's analysis. | Output Tokens: 5,906 Total number of tokens produced by the LLM (including reasoning/thinking tokens) for this paper's analysis.
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Finally, we evaluate our results using two real-world datasets and demonstrate that the proposed thresholds lead to lower losses than nonstrategic thresholds. |
| Researcher Affiliation | Academia | Aron Laszka and Yevgeniy Vorobeychik and Xenofon Koutsoukos Institute for Software Integrated Systems Department of Electrical Engineering and Computer Science Vanderbilt University Nashville, TN |
| Pseudocode | Yes | Theorem 2. Suppose that we are given a constant Λ, and the defender s choice is restricted to strategies where maxu A fu Lu Λ and minu A fu Lu Λ for a best response A2. Then, the output of the following algorithm is an optimal defense strategy: 1. For each user u, compute the loss of user u when it is not targeted as follows: ... 7. Output the strategy f. |
| Open Source Code | No | The paper does not provide concrete access to its own source code, nor does it explicitly state that the code is publicly available. |
| Open Datasets | Yes | UCI The first dataset is from the UCI Machine Learning Repository (Bache and Lichman 2013)... Enron The second dataset is the Enron e-mail dataset3 (Klimt and Yang 2004). 3http://www.cs.cmu.edu/ ./enron/ |
| Dataset Splits | No | The paper specifies training and testing splits (80% training, 20% testing for UCI; 12,000 training, 1,500 testing for Enron) but does not mention a separate validation split. |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., GPU/CPU models, memory specifications) used for running the experiments. |
| Software Dependencies | No | The paper mentions training a 'na ıve Bayes classifier' but does not provide specific version numbers for any software, libraries, or dependencies used in the experiments. |
| Experiment Setup | Yes | For every user, potential losses due to undelivered non-malicious and delivered targeted malicious e-mails are approximately ten times higher than losses due to delivered non-targeted e-mails. Formally, for each user u, Lu, Cu 10 Nu. The potential damage values Lu, Cu, and Nu follow a power law distribution. Formally, the number of users with damage values between some l and l + 1 is approximately twice as much as the number of users with values between l + 1 and l + 2. Finally, the value of Lu ranges from 0.5 to 5.5. |