The Theory of Kamusi
Kamusi is different from other dictionaries. Our goal, "Every Word in Every Language," requires a new way of thinking about how words, meanings, and translations relate to each other. Below are articles explaining the unique scientific precepts that guide our work.
Molecular Lexicography
Our data structure treats language as a matrix of interrelated data elements, linking terms at conceptual and functional nodes across languages, time, and space.
Concept Alignment
Instead of just matching spellings, we align concepts. This allows us to link 'light' (not heavy) in English to its true equivalent in Swahili, without confusing it with 'light' (not dark).
Party Terms (MWEs)
Multi-word expressions like 'kick the bucket' are treated as unique entries, allowing for accurate translation of idioms and phrases that stump other systems.
A New Solution for Dialects
We overcome the obstacles of defining boundaries between dialects by associating terms with geographic ranges instead of fixed, standard groups.
Knowledge from Crowds
Through games and targeted microtasks, we learn language directly from the public, turning the knowledge in people's heads into structured, shareable data.
Revolutionizing Machine Translation
By focusing on a high-quality, concept-based lexicon, our data provides the foundation for a quantum leap in the accuracy of automated translation.