ultk.effcomm
Tools for measuring languages for communicative efficiency.
Submodules divide the labor of a computational experiment performing an efficiency analysis of a language into several parts: generating and sampling the space of possible languages, measuring their properties, and determining which languages optimize efficient trade-offs w.r.t these properties.
The altk.effcomm.sampling submodule implements several methods for generating hypothetically possible languages of a given type, by sampling from a set of possible expressions, or permuting the expression-meaning mapping of an existing language.
The altk.effcomm.optimization submodule contains a general implementation of an evolutionary algorithm, which can be used to estimate a Pareto frontier of optimal solutions to an efficiency trade-off. It can also be used as a technique for randomly exploring the space of possible languages.
The altk.effcomm.tradeoff submodule contains tools for measuring a pool of languages for various properties, finding which languages are Pareto dominant with respect to two properties, and setting attributes of the language objects for further analysis.
The altk.effcomm.analysis submodule contains tools for performing numerical analyses and producing paradigmatic plots of languages in 2D trade-off space.
The altk.effcomm.rate_distortion submodule contains tools for information theory based analyses of the communicative efficiency of languages. Specificially, it includes methods for Rate-Distortion style (including the Information Bottleneck) analyses.
The altk.effcomm.agent submodule implements classes for constructing various speakers and listeners of a language. These are typically used in static analyses of informativity of a language, and are unified abstractions from the Rational Speech Act framework. They can also be used for dynamic analyses, however (see the signaling game example).
The altk.effcomm.informativity submodule implements tools for computing the literal or pragmatic informativity of a language, based on speaker/listener abstractions described above.
1"""Tools for measuring languages for communicative efficiency. 2 3Submodules divide the labor of a computational experiment performing an efficiency analysis of a language into several parts: generating and sampling the space of possible languages, measuring their properties, and determining which languages optimize efficient trade-offs w.r.t these properties. 4 5The `altk.effcomm.sampling` submodule implements several methods for generating hypothetically possible languages of a given type, by sampling from a set of possible expressions, or permuting the expression-meaning mapping of an existing language. 6 7The `altk.effcomm.optimization` submodule contains a general implementation of an evolutionary algorithm, which can be used to estimate a Pareto frontier of optimal solutions to an efficiency trade-off. It can also be used as a technique for randomly exploring the space of possible languages. 8 9The `altk.effcomm.tradeoff` submodule contains tools for measuring a pool of languages for various properties, finding which languages are Pareto dominant with respect to two properties, and setting attributes of the language objects for further analysis. 10 11The `altk.effcomm.analysis` submodule contains tools for performing numerical analyses and producing paradigmatic plots of languages in 2D trade-off space. 12 13The `altk.effcomm.rate_distortion` submodule contains tools for information theory based analyses of the communicative efficiency of languages. Specificially, it includes methods for Rate-Distortion style (including the Information Bottleneck) analyses. 14 15The `altk.effcomm.agent` submodule implements classes for constructing various speakers and listeners of a language. These are typically used in static analyses of informativity of a language, and are unified abstractions from the Rational Speech Act framework. They can also be used for dynamic analyses, however (see the [signaling game example](https://clmbr.shane.st/altk/src/examples/signaling_game)). 16 17The `altk.effcomm.informativity` submodule implements tools for computing the literal or pragmatic informativity of a language, based on speaker/listener abstractions described above. 18"""