Policy Reasoning & Secure Information Exchange
A policy is something a human wrote in words. Every system that enforces one has to get those words into a machine, and the usual method is to paraphrase them into code — at which point the sentence the accountable human signed and the behavior the machine actually exhibits are two different artifacts, and only one of them is running. Our answer is to compile the sentence: SBVR Structured English in, BVR and OWL 2 RL out, BaseVISor executing, with the original sentence still readable by the officer who owns it. The payoff is procurement-grade, and it is the CogSIMA paper’s, not our marketing department’s: “mechanisms can be accredited and then policies can be added dynamically.” Accredit the mechanism once; change the policy as data.
The Challenge: There Is Nowhere to Put the Guard
Two organizations have to exchange information. Each has its own policy, written by its own authority, and neither can impose that policy on the other. This is the ordinary condition of a coalition, a joint task force, or any two agencies that talk — and it is the condition under which the standard answer, put a guard in the middle, has no middle to be put in.
Our flagship paper takes that condition to its sharpest instance: XMPP, the IETF-standardized messaging and presence protocol that carries chat across defense and financial systems. XMPP is federated by design. Every user is hosted on his own server; servers speak to servers; a message from one domain to another crosses two independently administered machines, each with its own policy authority and no power over the other. There is no chokepoint. The paper states the consequence in a single sentence, and the whole page hangs on it:
“The distributed nature of XMPP makes centralized information exchange policy enforcement impossible, however.”
Nor does the protocol help. XMPP gives a user privacy lists — block this contact — but it gives the server almost nothing: “However, there are currently no methods for server to specify policies to restrict users’ chat, except by name.” A name-based blocklist. You can say no chat with this account; you cannot say no chat about classified material with a foreign national on this channel. That gap is the entire problem, and it is not peculiar to chat — it is what every federated exchange looks like from the inside.
Our Approach: Compile the Sentence
A policy is something a human wrote in words. VIStology’s PolVISor pipeline compiles the words instead of paraphrasing them: a policy authored in SBVR Structured English — an OMG standard a person can read back — is translated automatically into executable rules (BVR plus OWL 2 RL) and enforced by a BaseVISor reasoner at every server in a federated XMPP system, where no central guard is possible. A prohibited message is dropped before the server sees it, and every decision carries a machine-readable explanation.
When two organizations each bring their own policy, the policies themselves are reconciled — in the strongest result, even when the two sides describe the world in different (but similar) vocabularies. And because a policy about what you may believe is still a policy, the same machinery graded open-source reports against NATO STANAG 2022. No paper in this cluster reports a performance evaluation; each use-case page says so, in its own words.
Use Cases
Policy-Governed Information Exchange
Structured English in, executable rules out: information-exchange policies compiled and enforced on every message in federated XMPP — with the reasoner’s explanation behind each decision.
Learn more →Security Policy Reconciliation
Two policies, one exchange: reconciliation across independently administered servers, demonstrated even across different (but similar) vocabularies.
Learn more →Proof Points
No paper in this cluster carries a number. What they carry is provenance, architecture and a working system — and each one is labeled for what it is, because every claim below can be checked against a published PDF or a federal spending record.
A Million-Dollar Army Contract, Named “Security Policy Reconciliation”
PolVISor was built under US Army SBIR Phase II contract W91260-09-C-0037, in the STIDS paper’s own acknowledgment: “This work was performed under US Army contract W91260-09-C-0037 ‘Security Policy Reconciliation’.” The sibling CogSIMA paper names the command and a second contract: “This research was performed under U.S. Army contracts W91260-09-C-0037 (USASMDC) and W15P7T-09-C-S031 (CERDEC I2WD).” The award is independently confirmed on USASpending — SBIR Phase II, $999,995, 20 May 2009 to 20 May 2011. Boundary, stated: an SBIR Phase II is a research contract. There was no transition, no fielding and no program of record, and PolVISor is a research prototype, not a product.
One Policy, One Closed World
Presented at IEEE POLICY 2008 — the field’s canonical policy venue — an extended BaseVISor that reasons across several policies at once: three procedural attachments implementing N3’s log:semantics, log:includes and log:notIncludes, so that each policy is materialized into its own closed-world fact base and bound to a rule variable. “This work was partially funded by the United States Missile Defense Agency under contract number W9113M-08-C-0053.” Boundaries, because the title is bait: it is a three-page demonstration abstract in the future tense (“This demonstration will show…”) and it never reports the demo ran; the extension is not deontic logic; there is no conflict detection in it; and the policy ontology and both scenarios are other researchers’ work, adopted with credit.
Selected Publications
The peer-reviewed record behind VIStology’s policy-reasoning practice. Read them — the limits on this page are the limits in these papers.
Inference and Ontologies
In A. Kott, C. Wang & R.F. Erbacher (Eds.), Cyber Defense and Situational Awareness, Advances in Information Security 62, Springer, pp. 167-199 (2014)
View PublicationA Demonstration of Formal Policy Reasoning Using an Extended Version of BaseVISor
2008 IEEE Workshop on Policies for Distributed Systems and Networks (POLICY 2008), pp. 225-227
View PublicationSemantic Policy Enforcement and Reconciliation for Information Exchange in XMPP
Proceedings of the Sixth International Conference on Semantic Technologies for Intelligence, Defense, and Security (STIDS 2011), CEUR-WS Vol-808, pp. 22-29
View PDFSituations and Policies
2012 IEEE International Multi-Disciplinary Conference on Cognitive Methods in Situation Awareness and Decision Support (CogSIMA), pp. 7-14
View PublicationToward Formal Reasoning with Epistemic Policies about Information Quality in the Twittersphere
Proceedings of the 14th International Conference on Information Fusion (FUSION 2011), Chicago, IL, pp. 1271-1278, ISIF
View PublicationProvenance and its limits, stated once. The PolVISor work is cleanly VIStology’s: four of the five STIDS authors publish under VIStology, Inc., and it was funded by US Army SBIR Phase II W91260-09-C-0037 (USASMDC), “Security Policy Reconciliation” — confirmed on USASpending at $999,995, May 2009 to May 2011. CogSIMA names a second Army contract, W15P7T-09-C-S031 (CERDEC I2WD), and carries a U.S. Government co-author: Gerald M. Powell, U.S. Army Research Laboratory, Aberdeen Proving Ground. POLICY 2008 was “partially funded by the United States Missile Defense Agency under contract number W9113M-08-C-0053.” Two papers carry no sponsor at all, and we say so rather than letting the neighbors’ funding rub off on them: the Springer chapter has no funding statement of any kind (its volume was edited by Army researchers — that is a fact about the book, not about who paid for the chapter), and the FUSION paper names no sponsor. None of the five reports an evaluation.
Related
Take the Policy Out of the Code
If your exchange rules live inside a program — a guard, a filter, a broker, a branch someone wrote in 2014 — then every change to the policy is a change to software, and the sentence your accreditor signed is not the one the machine is running. We can compile the sentence instead.
Get in Touch