The paper tackles a problem in probability theory that arises when you have incomplete information, not just uncertainty but genuine gaps in evidence where you simply do not know something. Standard probability theory assumes every event either happens or does not happen, but real situations sometimes involve missing data where neither option can be confirmed. The authors use a branch of non-classical logic called paracomplete logic, specifically Logics of Formal Undeterminedness, to build a framework where probability can be assigned even when evidence is absent rather than just uncertain.
The core technical contributions are paracomplete versions of two foundational results in probability. The first is the law of total probability, which normally lets you break a complex probability calculation into simpler parts by considering all possible cases. The second is Bayes' rule, which lets you update your beliefs about a cause after observing an effect. Both results are reformulated to handle a three-valued situation where statements can be true, false, or undetermined, meaning the usual binary logic underlying standard probability no longer applies.
The practical upshot is a new kind of probability space designed for situations where data is missing rather than merely noisy or random. This matters for fields like artificial intelligence, legal reasoning, medical diagnosis, and database theory, where systems routinely face gaps in available information. By giving probabilities a rigorous logical foundation that acknowledges undeterminedness as a first-class concept rather than a workaround, the framework offers a more honest and potentially more useful tool for reasoning under incomplete evidence.