Towards Legal Contract Formalization with Controlled Natural Language Templates
Bibliographic record
Abstract
Automated formalization of legal texts in order to remove ambiguities, conflicts and incompleteness has been a challenge for Requirements Engineering (RE) research for decades. This work seeks to make an incremental step towards this objective for legal contracts by making use of contract templates. Our proposed approach starts with a natural language contract template, together with a manually formalized specification of that template. A contract writer can make customizations to the template, which trigger the automatic formalization of the corresponding customized contract. Our target specification language is Symboleo, which is created specifically for contract verification and monitoring. Starting with a manually formalized template reduces the complexity associated with a fully automated formalization. Typical contract templates use simple fill-in-the-blank parameters, which serve as customizations to formalize in our framing of the problem. Our approach pushes the boundaries of these templates by allowing the contract writer to enter complex natural language customizations, such as prepositional phrases and conditional statements. This work explores what types of natural language patterns can be used in that context by analyzing relevant linguistics and real legal contracts. It also introduces a tool, SymboleoNLP, that suggests the feasibility of the formalization process.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.016 | 0.035 |
| Meta-epidemiology (narrow) | 0.001 | 0.002 |
| Meta-epidemiology (broad) | 0.001 | 0.004 |
| Bibliometrics | 0.003 | 0.002 |
| Science and technology studies | 0.002 | 0.006 |
| Scholarly communication | 0.007 | 0.009 |
| Open science | 0.004 | 0.004 |
| Research integrity | 0.002 | 0.005 |
| Insufficient payload (model declined to judge) | 0.005 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".