Language & meaning
How do words, context and linguistic structure contribute to understanding? We investigate relevance at the level of individual word occurrences and the evidence available within a language.
Semantics · Context · Word relevanceLanguage.
Reasoning.
Representation.
And Building.
Exploring how meaning takes shape.
Turning theory into things that work.
01 / Our purpose
Big questions.
Concrete contributions.
Langvos investigates language as a way to explore how meaning, reasoning and representations emerge and relate.
Computational linguistics is our working ground: a place to formulate original ideas, make them testable, and develop applications that address real needs.
Our mission
Advance the understanding of language, reasoning and representation through original computational research, and transform that knowledge into practical methods, tools and solutions.
Our vision
Build a research environment where ambitious ideas about language become rigorous investigations and useful technologies—connecting scientific curiosity with real-world impact.
A philosophical horizon
What if language offers a broader way to look at the world?
This question guides our curiosity. It is a starting point for inquiry, not an established scientific claim. Interaction, information and language are distinct concepts; understanding their relationships requires precise definitions and evidence.
Our task is to turn parts of that broad perspective into specific questions that can be investigated, tested and challenged.
02 / Research directions
Questions worth
working on.
Our initial agenda connects foundational questions with experimental methods and applied engineering.
Reasoning & representation
How can representations of language support inference and understanding? We explore how information is organized, which distinctions a representation preserves, and how those choices affect reasoning.
Knowledge · Structure · InferenceLearning from linguistic evidence
What can be learned from a corpus when dictionaries, translations and pretrained models are unavailable? We study computational approaches under explicit information constraints.
Corpus-based methods · Uncertainty · InformationEvaluation & applied systems
How do we test ideas rigorously and turn them into useful systems? We develop benchmarks, experimental workflows and software that connect research questions to reproducible evidence and practical applications.
Benchmarks · Human judgment · Research engineering
Understanding the world through evidence, data and behavior.
Langvos aims to build models that use language to represent the world, reason about it and turn understanding into practical solutions.
Research in development
The Popkorn project
Which words matter most
for understanding a sentence?
Popkorn investigates how to rank word occurrences by their contribution to sentence understanding in an unknown language, using only a corpus in that language.
The project brings together corpus-based estimation, benchmark construction and human judgments to examine this question experimentally.
Explore the Popkorn project03 / People
Research guided
by a builder’s mindset.
Coordinator & Principal Investigator
Luiz Carlos d’Oleron
Computer scientist, engineer and author.
With more than two decades of programming experience and a background in real-time and industrial systems, Luiz brings a practical engineering perspective to computational linguistics.
At Langvos, he connects fundamental questions about language with the discipline of designing, implementing and evaluating working systems.
Ideas to question.
Knowledge to build with.