I am trying to make a dataset based on MITRE ATT&CK (https://attack.mitre.org/) website to create AI-based threat intelligence. However, I didn't find any automated tools or readymade datasets for it.
Is there any dataset based on the MITRE ATT&CK?
Thanks in advance.
Unfortunately, to my knowledge, there are not. I think this is probably, at least in part, because any readymade dataset is going to lack re-inforcement training of the data. Given that MITRE ATT&CK is going to be applied very differently depending on the type and nature of data in the dataset, I would expect the datasets to be relatively inaccurate, at least in most applications.
Very interesting research - this form of AI-based threat intelligence is needed. One thought I had is that you could utilize big data for this, and apply a form of regression analysis between known threats based on MITRE ATT&CK and known datasets with no threat. Feel free to reach out if you have interest in discussing further.
Hi Noah M. Kenney ,
Thank you for sharing such an informative suggestion. Do you have any particular dataset or research paper for this?
Thank you.
Anik Islam Abhi I am working on a relevant paper currently (happy to share findings offline). It covers a regression analysis and AI modeling of data to detect anomalies.
I suspect that another thing you'll need to deal with is how to do this threat analysis in a manner that does not lead to data privacy concerns for both those in the dataset and the data being analyzed for threat intelligence (i.e., ensuring anonymity). I do have a paper that addresses this. Although it is not directed toward your specific research question, it may be useful for that challenge of structuring your dataset: Article Differentially Private Data and Data De-Identification
building a quality dataset for AI-based threat intelligence requires careful curation, data cleaning, and validation. Ensure that you handle any collected data ethically, respecting privacy and legal considerations. Additionally, keep an eye on updates from MITRE and the cybersecurity community, as new tools or resources may become available over time to aid in dataset creation.
- Badges
- Science topic