--- title: "The Law of Inertia and the Frame Problem in Attempto Controlled" description: "The paper addresses the frame problem: how a logical description can state that whatever an action does not touch stays as it… 6 references." authors: ["Norbert E. Fuchs"] year: 2020 venue: "Proceedings of the Seventh International Workshop on Controlled Natural Language (CNL 2020/21)" source: "https://aclanthology.org/2021.cnl-1.7/" licence: "CC BY 4.0 (ACL Anthology)" licence_url: "https://creativecommons.org/licenses/by/4.0/" summary_licence: "Elamur Cosmic License" canonical: "https://research.elamur.ai/free/en/papers/fuchs-2021-law-inertia-frame-problem-attempto/" --- [Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/fuchs-2021-law-inertia-frame-problem-attempto/) [home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2020](https://md.research.elamur.ai/free/en/papers/2020/page/11/) / The Law of Inertia and the Frame Problem in Attempto Controlled English # The Law of Inertia and the Frame Problem in Attempto Controlled English **Authors:** [Norbert E. Fuchs](https://md.research.elamur.ai/free/en/authors/norbert-e-fuchs/) · **Year:** 2020 · **Venue:** [Proceedings of the Seventh International Workshop on Controlled Natural Language (CNL 2020/21)](https://md.research.elamur.ai/free/en/venues/cnl/) ([International Workshop on Controlled Natural Language 2020](https://md.research.elamur.ai/free/en/venues/cnl/2020/)) · **Publisher:** [Special Interest Group on Controlled Natural Language](https://md.research.elamur.ai/free/en/publishers/special-interest-group-on-controlled-natural-language/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/) **Links:** [Publisher's page](https://aclanthology.org/2021.cnl-1.7/) · [PDF](https://aclanthology.org/2021.cnl-1.7.pdf) ## Summary The paper addresses the frame problem: how a logical description can state that whatever an action does not touch stays as it was, which is Leibniz's "law of inertia". Fuchs recasts the Yale Shooting Problem as a harmless scenario with a wine glass and a bottle. A plain first-order encoding of it admits an unintended model in which the bottle spontaneously closes again. He then takes the default-logic treatment of inertia and rewrites it as an implication that uses negation as failure over time points. Next, he shows that this rule can be written directly as an Attempto Controlled English (ACE) sentence. The Attempto reasoner RACE proves the expected outcomes for a "continuous sleep" case and an "interrupted sleep" case. Writing one inertia sentence per fluent does not scale, so the rule is moved into RACE itself as generic Prolog auxiliary axioms. With these built-in axioms RACE returns only the intended solution of the glass-and-bottle problem. The result shows that commonsense persistence can be stated readably in a controlled language and used efficiently in reasoning. ## Key points - The Yale Shooting variant uses four time points (0–3) and two fluents, *empty* and *open*. Encoding only the changes yields two consistent valuations, and only one of them is intended. - The inertia rule reads, roughly: if X holds at T1, T2 is later, and it cannot be proved that X fails at T2, then X holds at T2. Its ACE form is a single English-like conditional. - RACE relies on about 100 auxiliary Prolog axioms, covering things like plural/singular links and number relations. Each one is invoked only when a proof needs it, so the large number costs little. - In its basic form, RACE cannot handle logical negation inside negation as failure. The workaround is to swap "does not sleep" for "wakes". - Inertia is built into RACE as four frame axioms: - persistence of intransitive verbs; - persistence of transitive verbs; - persistence of copula plus adjective; - persistence of fluents that appear in the precondition of an implication. Ditransitive verbs are left out because they seem not to trigger the frame problem. As a side effect, the built-in axioms also lift the negation restriction. ## Interesting ideas - Persistence works as a default: a state carries forward to a later time unless something in the text proves it stopped. This lets a few explicit changes stand in for a large number of frame axioms. - Instead of a per-fluent inertia rule, generic axioms are keyed by syntactic shape: intransitive verb, transitive verb, and copula+adjective. Shape matters because each one has a different internal representation. - A dedicated frame axiom links persisted fluents to the preconditions of conditional rules, so that later implications fire on states inherited from earlier times. - RACE's output lists the minimal subset of axioms needed for a proof, which serves as a built-in explanation. - The paper notes that inertia only projects forward in time, and that reasoning backwards needs a different mechanism such as abduction. ## For Mova How these ideas could serve Mova, our small open language model — its files, stages and tests: - **I1** (world): The world-of-text state tracker could copy every entity state and holding forward along the event order, and drop it only when the text explicitly asserts a change or negation. - **I2** (frames): Frame-to-state mapping could use a small set of persistence templates indexed by argument structure, rather than hand-coding a rule for each VerbNet class. - **I3** (world): When a story states a rule such as "if the glass is empty and the bottle is open…", the reasoner should check it against inherited states as well as states mentioned at that exact moment. - **I4** (answer): Each answer to a fable or school-test question could return the minimal set of sentences and world facts used, which supports the explainability requirement. - **I5** (answer): Questions like "why was the glass empty before?" should be routed to a separate abductive step rather than to the forward persistence logic, which avoids unsound backward inferences. *Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.* 📚 Shelves (2): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 18,614 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 17,965 works 🧭 Topics (1): [English](https://md.research.elamur.ai/free/en/topics/english/) · 2,379 works ## Cite Cite as: Norbert E. Fuchs. 2020. The Law of Inertia and the Frame Problem in Attempto Controlled English. Proceedings of the Seventh International Workshop on Controlled Natural Language (CNL 2020/21). The paper itself, not this page. BibTeX from the library's record of the work. ```bibtex @inproceedings{fuchs2020law, title = {{The Law of Inertia and the Frame Problem in Attempto Controlled English}}, author = {Norbert E. Fuchs}, booktitle = {Proceedings of the Seventh International Workshop on Controlled Natural Language (CNL 2020/21)}, year = {2020}, url = {https://aclanthology.org/2021.cnl-1.7/} } ``` - [← previous The Biblical Hebrew infinitive](https://md.research.elamur.ai/free/en/papers/doron-2020-biblical-hebrew-infinitive/) - [next → The Political Reform and the Amharic Biweekly Reporter: The Editorial Contents of the Reporter Newspaper in Ethiopia from 2017 to 2019](https://md.research.elamur.ai/free/en/papers/gessese-2020-political-reform-amharic-biweekly-reporter/) - [↑ back to papers of 2020](https://md.research.elamur.ai/free/en/papers/2020/page/11/) ## References 6 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet. 1. E. Erdem, M. Gelfond and N. Leone. 2016. Applications of Answer Set Programming. ACM. (www.depts.ttu.edu/cs/research/documents/5.pdf) [find](https://scholar.google.com/scholar?q=E%2E+Erdem%2C+M%2E+Gelfond+and+N%2E+Leone%2E+2016%2E+Applications+of+Answer+Set+Programming%2E+ACM%2E+%28www%2Edepts%2Ettu%2Eedu%2Fcs%2Fresearch%2Fdocuments%2F5%2Epdf%29) 2. N. E. Fuchs. 2012. First-Order Reasoning for Attempto Controlled English. In Proc. of the Second International Workshop on Controlled Natural Language (CNL 2010). Marettimo, Italy. [find](https://scholar.google.com/scholar?q=N%2E+E%2E+Fuchs%2E+2012%2E+First%2DOrder+Reasoning+for+Attempto+Controlled+English%2E+In+Proc%2E+of+the+Second+International+Workshop+on+Controlled+Natural+Language+%28CNL+2010%29%2E+Marettimo%2C+Italy%2E) 3. N. E. Fuchs. 2016. Reasoning in Attempto Controlled English: Non-monotonicity. In Proc. of the Fifth International Workshop on Controlled Natural Language (CNL 2016). Aberdeen, UK . [find](https://scholar.google.com/scholar?q=N%2E+E%2E+Fuchs%2E+2016%2E+Reasoning+in+Attempto+Controlled+English%3A+Non%2Dmonotonicity%2E+In+Proc%2E+of+the+Fifth+International+Workshop+on+Controlled+Natural+Language+%28CNL+2016%29%2E+Aberdeen%2C+UK+%2E) 4. S. Hanks and D. McDermott. 1987. Nonmonotonic logic and temporal projection. Artificial Intelligence, 33(3):379– 412. [find](https://scholar.google.com/scholar?q=S%2E+Hanks+and+D%2E+McDermott%2E+1987%2E+Nonmonotonic+logic+and+temporal+projection%2E+Artificial+Intelligence%2C+33%283%29%3A379%E2%80%93+412%2E) 5. R. Kowalski and M. Sergot. 1986. A logic-based calculus of events. New Generation Computing, 4 (1): 67–95 (1986). G. Leibniz. 1679. An Introduction to a Secret Encyclopaedia. I. Newton. 1687. Philosophiæ Naturalis Principia Mathematica. [find](https://scholar.google.com/scholar?q=R%2E+Kowalski+and+M%2E+Sergot%2E+1986%2E+A+logic%2Dbased+calculus+of+events%2E+New+Generation+Computing%2C+4+%281%29%3A+67%E2%80%9395+%281986%29%2E+G%2E+Leibniz%2E+1679%2E+An+Introduction+to+a+Secret+Encyclopaedia%2E+I%2E+Newton%2E+1687%2E+Philosop) 6. R. Reiter. 1980. A logic for default reasoning. Artificial Intelligence, 13:81-132 (1980). [find](https://scholar.google.com/scholar?q=R%2E+Reiter%2E+1980%2E+A+logic+for+default+reasoning%2E+Artificial+Intelligence%2C+13%3A81%2D132+%281980%29%2E) *Updated 2 November 2026*