Riesz Logic

Daoud Clarke
DOI: https://doi.org/10.48550/arXiv.1410.2910
2014-10-11
Abstract:We introduce Riesz Logic, whose models are abelian lattice ordered groups, which generalise Riesz spaces (vector lattices), and show soundness and completeness. Our motivation is to provide a logic for distributional semantics of natural language, where words are typically represented as elements of a vector space whose dimensions correspond to contexts in which words may occur. This basis provides a lattice ordering on the space, and this ordering may be interpreted as "distributional entailment". Several axioms of Riesz Logic are familiar from Basic Fuzzy Logic, and we show how the models of these two logics may be related; Riesz Logic may thus be considered a new fuzzy logic. In addition to applications in natural language processing, there is potential for applying the theory to neuro-fuzzy systems.
Logic in Computer Science,Computation and Language
What problem does this paper attempt to address?
The main problem that this paper attempts to solve is to combine distributional semantics with logical semantics and provide a new logical framework for natural language processing. Specifically, the author introduced Riesz Logic (RL), whose models are abelian lattice - ordered groups, and these models generalize Riesz spaces (vector lattices). In this way, the author hopes to bridge the gap between distributional semantics and logical semantics in natural language processing, thus opening up new research areas and problem - solving methods. ### Detailed description of the main problem 1. **Combining distributional semantics and logical semantics**: - Distributional semantics usually uses vector spaces to represent the meanings of words, where each dimension corresponds to different contexts in which the word appears. - Logical semantics is more concerned with how to translate sentences into logical forms. - The goal of the paper is to combine these two methods through Riesz Logic so that the statistical characteristics of distributional semantics and the formal reasoning ability of logical semantics can be utilized simultaneously. 2. **Defining a new logical system**: - The models of Riesz Logic are abelian lattice - ordered groups, which are a special type of vector lattices. - The author defined a series of axioms and inference rules to ensure the rationality and completeness of this new logical system. 3. **Applying to neuro - fuzzy systems**: - Besides natural language processing, the author also explored the potential applications of Riesz Logic in neuro - fuzzy systems. - In particular, they discussed how to combine Riesz Logic with neural networks in deep learning, such as using the ReLU activation function. ### Markdown representation of formulas - **Definition of partially ordered vector spaces**: \[ \text{Definition 1 (partially ordered vector space)}: V \text{ is a real vector space, and there is a partial order relation } \leq \text{ satisfying:} \] \[ \text{If } u \leq v \text{ then } u + w \leq v + w \] \[ \text{If } u \leq v \text{ then } \alpha u \leq \alpha v, \forall \alpha \geq 0 \] - **Interpretation of RL**: \[ J\varphi \to \psi K = J\psi K - J\varphi K \] \[ \text{If } u \leq v \text{ then } \alpha u \leq \alpha v, \forall \alpha \geq 0 \] \[ J0K = 0 \] Through these definitions and formulas, the author demonstrates the rationality and completeness of Riesz Logic and provides a theoretical basis for its applications in natural language processing and other fields.