Pojemno

Pojemno is a hypothetical term used in discussions of information representation and linguistics to describe the expressive capacity of a medium to encode distinct concepts. The term is not part of standard jargon in major disciplines and is used here as a tool for comparing encoding schemes.

Pojemno refers to the maximum number of distinct concepts that can be encoded and later retrieved with a predefined level of disambiguation within a medium, such as text, a dataset, or a language model, under constraints like noise and encoding length.

Pojemno is typically assessed by experiments that vary the number of target concepts and measure retrieval accuracy, confusion rates, or cross-entropy between intended and recovered concepts. It can be reported as an index or as a dimensionless ratio that supports comparisons under similar conditions.

In practice, pojemno serves as a metric for vocabulary design, ontology granularity, and capacity planning in models and datasets. It helps evaluate trade-offs between conciseness and expressiveness and can guide decisions about encoding schemes, such as discrete symbol vocabularies versus distributed representations.

An example: a corpus with high pojemno allows distinct concepts to be identified with minimal context, while a low-pojemno corpus tends to require more context or yields greater ambiguity.

The applicability of pojemno depends on the encoding method, concept definitions, and acceptable error thresholds; comparisons should align these factors to be meaningful.