Believable Fake Scientific Document Generation by Exploring Data Semantics
Abstract
Statement of Work:In this work, the PI will investigate methods and techniques and develop tools toward automatically generating believable fake scientific documents that can help reduce damage to an organization~s intellectual property and assets. The key scientific innovations of this proposal are the following:~ A mathematical notion of syntactic believability,~ A mathematical notion of semantic believability,~ Algorithms to identify parts of a document that can be modified with false information,~ Semantic falsification methods to modify documents, and~ Algorithms to create optimal fake documents.Objective:The objective for this research is to study and develop viable techniques and tools for automatically generatingbelievable fake scientific documents that can help reduce damage to an organization~s intellectual property and assets by: (i) creating costs associated information overload by forcing intruder to wade through many irrelevant and insignificant documents, and (ii) by explicitly planting fake scientific documents that mislead the intruder over time.Approach:In this proposal the PI will develop methods and techniques and develop tools toward automatically generating believable fake scientific documents that can help reduce damage to an organization~s intellectual property and assets.For a fake document d to mislead an adversary, the document d must have several properties that make it believableto an adversary.~ First, d must be syntactically believable. An attacker will have no problem identifying true documents d~ on the topic tof interest to him. Any fake document d~ must have a syntactic form that makes is similar to this plethora of documents.~ Second, d must be semantically believable. In particular, d must look sufficiently credible to an expert in the field oftopic t that the expert is forced to spend a certain amount of time looking at the design shown in document d.The key scientific innovations of this proposal are therefore the following.1T. ASK 1 Developing a mathematical notion of syntactic believability.2T. ASK 2 Developing a mathematical notion of semantic believability.3T. ASK 3 Developing structural pattern mining algorithms to identify parts of a document that can be modified withfalse information.4T. ASK 4 Developing a suite of semantic falsification methods to modify documents.5T. ASK 5 Optimally creating a fake document.Overall Merit and ONR Mission/Relevance:Adversaries have been covertly gathering Navy~s information (digital artifacts). Successful results from this researchwill pave a way for diluting adversary understanding and confidence on digital artifacts they collected, shifting awayand eroding adversaries~ knowledge regarding Navy~s assets, and hence shaping adversary~s situational awareness toNavy~s advantage.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Sep 30, 2016
- Source ID
- N000141612896
Entities
People
- Sushil Jajodia
Organizations
- George Mason University
- Office of Naval Research
- United States Navy