scholarly journals Data-driven Conceptual Spaces: Creating Semantic Representations For Linguistic Descriptions Of Numerical Data

2018 ◽  
Vol 63 ◽  
pp. 691-742 ◽  
Author(s):  
Hadi Banaee ◽  
Erik Schaffernicht ◽  
Amy Loutfi
Keyword(s):  
A Priori ◽  
Empirical Evaluation ◽  
Numerical Data ◽  
Semantic Interpretation ◽  
Data Driven ◽  
Conceptual Spaces ◽  
Semantic Features ◽  
Semantic Representations ◽  
Quality Dimensions ◽  
Using Data

There is an increasing need to derive semantics from real-world observations to facilitate natural information sharing between machine and human. Conceptual spaces theory is a possible approach and has been proposed as mid-level representation between symbolic and sub-symbolic representations, whereby concepts are represented in a geometrical space that is characterised by a number of quality dimensions. Currently, much of the work has demonstrated how conceptual spaces are created in a knowledge-driven manner, relying on prior knowledge to form concepts and identify quality dimensions. This paper presents a method to create semantic representations using data-driven conceptual spaces which are then used to derive linguistic descriptions of numerical data. Our contribution is a principled approach to automatically construct a conceptual space from a set of known observations wherein the quality dimensions and domains are not known a priori. This novelty of the approach is the ability to select and group semantic features to discriminate between concepts in a data-driven manner while preserving the semantic interpretation that is needed to infer linguistic descriptions for interaction with humans. Two data sets representing leaf images and time series signals are used to evaluate the method. An empirical evaluation for each case study assesses how well linguistic descriptions generated from the conceptual spaces identify unknown observations. Furthermore, comparisons are made with descriptions derived on alternative approaches for generating semantic models.

scholarly journals Hierarchical Linear Disentanglement of Data-Driven Conceptual Spaces

2020 ◽  
Author(s):  
Rana Alshaikh ◽  
Zied Bouraoui ◽  
Steven Schockaert
Keyword(s):  
Vector Space ◽  
Representation Learning ◽  
Data Driven ◽  
Vector Spaces ◽  
Conceptual Spaces ◽  
Semantic Features ◽  
Geometric Meaning ◽  
Domain Specific ◽  
Text Corpora ◽  
Quality Dimensions

Conceptual spaces are geometric meaning representations in which similar entities are represented by similar vectors. They are widely used in cognitive science, but there has been relatively little work on learning such representations from data. In particular, while standard representation learning methods can be used to induce vector space embeddings from text corpora, these differ from conceptual spaces in two crucial ways. First, the dimensions of a conceptual space correspond to salient semantic features, known as quality dimensions, whereas the dimensions of learned vector space embeddings typically lack any clear interpretation. This has been partially addressed in previous work, which has shown that it is possible to identify directions in learned vector spaces which capture semantic features. Second, conceptual spaces are normally organised into a set of domains, each of which is associated with a separate vector space. In contrast, learned embeddings represent all entities in a single vector space. Our hypothesis in this paper is that such single-space representations are sub-optimal for learning quality dimensions, due to the fact that semantic features are often only relevant to a subset of the entities. We show that this issue can be mitigated by identifying features in a hierarchical fashion. Intuitively, the top-level features split the vector space into different domains, making it possible to subsequently identify domain-specific quality dimensions.

scholarly journals Atmospheric CO<sub>2</sub> inversions on the mesoscale using data-driven prior uncertainties: quantification of the European terrestrial CO<sub>2</sub> fluxes

2018 ◽  
Vol 18 (4) ◽  
pp. 3047-3064 ◽  
Author(s):  
Panagiotis Kountouris ◽  
Christoph Gerbig ◽  
Christian Rödenbeck ◽  
Ute Karstens ◽  
Thomas F. Koch ◽  
...  
Keyword(s):  
Spatial Representation ◽  
A Priori ◽  
Regional Scale ◽  
Data Driven ◽  
Biogenic Carbon ◽  
Sensitivity Tests ◽  
Uncertainty Estimates ◽  
Using Data ◽  
A Priori Uncertainty ◽  
Photosynthesis And Respiration

Abstract. Optimized biogenic carbon fluxes for Europe were estimated from high-resolution regional-scale inversions, utilizing atmospheric CO2 measurements at 16 stations for the year 2007. Additional sensitivity tests with different data-driven error structures were performed. As the atmospheric network is rather sparse and consequently contains large spatial gaps, we use a priori biospheric fluxes to further constrain the inversions. The biospheric fluxes were simulated by the Vegetation Photosynthesis and Respiration Model (VPRM) at a resolution of 0.1° and optimized against eddy covariance data. Overall we estimate an a priori uncertainty of 0.54 GtC yr−1 related to the poor spatial representation between the biospheric model and the ecosystem sites. The sink estimated from the atmospheric inversions for the area of Europe (as represented in the model domain) ranges between 0.23 and 0.38 GtC yr−1 (0.39 and 0.71 GtC yr−1 up-scaled to geographical Europe). This is within the range of posterior flux uncertainty estimates of previous studies using ground-based observations.

scholarly journals Atmospheric CO<sub>2</sub> inversions at the mesoscale using data driven prior uncertainties. Part2: the European terrestrial CO<sub>2</sub> fluxes

2016 ◽  
Author(s):  
Panagiotis Kountouris ◽  
Christoph Gerbig ◽  
Christian Rödenbeck ◽  
Ute Karstens ◽  
Thomas F. Koch ◽  
...  
Keyword(s):  
Spatial Representation ◽  
A Priori ◽  
Regional Scale ◽  
Data Driven ◽  
Biogenic Carbon ◽  
Sensitivity Tests ◽  
Uncertainty Estimates ◽  
Using Data ◽  
A Priori Uncertainty ◽  
Photosynthesis And Respiration

Abstract. Optimized biogenic carbon fluxes for Europe were estimated from high resolution regional scale inversions, utilizing atmospheric CO2 measurements at 16 stations for the year 2007. Additional sensitivity tests with different data-driven error structures were performed. As the atmospheric network is rather sparse and consequently contains large spatial gaps, we use a priori biospheric fluxes to further constrain the inversions. The biospheric fluxes were simulated by the Vegetation Photosynthesis and Respiration Model (VPRM) at a resolution of 0.1° and optimized against Eddy covariance data. Overall we estimate an a priori uncertainty of 0.54 GtC y−1 related to the poor spatial representation between the biospheric model and the ecosystem sites. The sink estimated from the atmospheric inversions for the area of Europe (as represented in the model domain) ranges between 0.23 and 0.38 GtC y−1 (0.30 and 0.49 GtC y−1 up-scaled to geographical Europe). This is within the range of posterior flux uncertainty estimates of previous studies using ground based observations.

scholarly journals Implicatures and Naturalness

2021 ◽  
pp. 143-163
Author(s):  
Igor Douven
Keyword(s):  
A Priori ◽  
Conceptual Spaces ◽  
Using Data ◽  
Independent Reality

AbstractPragmatics postulates a rich typology of implicatures to explain how true assertions can nevertheless be misleading. This typology has been mainly defended on the basis of a priori considerations. We consider the question of whether the typology corresponds to an independent reality, specifically whether the various types of implicatures constitute natural concepts. To answer this question, we rely on the conceptual spaces framework, which represents concepts geometrically, and which provides a formally precise criterion for naturalness. Using data from a previous study, a space for the representation of implicatures is constructed. Examination of the properties of various types of implicatures as represented in that space then gives some reason to believe that most or even all types of implicatures do correspond to natural concepts.

Development of a Pediatric Early Warning System Using Data-Driven Vital Signs

PEDIATRICS ◽  
2016 ◽  
Vol 137 (Supplement 3) ◽  
pp. 256A-256A
Author(s):  
Catherine Ross ◽  
Iliana Harrysson ◽  
Lynda Knight ◽  
Veena Goel ◽  
Sarah Poole ◽  
...  
Keyword(s):  
Early Warning ◽  
Early Warning System ◽  
Vital Signs ◽  
Warning System ◽  
Data Driven ◽  
Using Data

Recognizing Job Apathy Patterns of Iraqi Higher Education Employees Using Data Mining Techniques

2019 ◽  
Vol 54 (4) ◽  
Author(s):  
Mustafa S. Abd ◽  
Suhad Faisal Behadili
Keyword(s):  
Higher Education ◽  
Data Mining ◽  
Psychiatric Patients ◽  
Human Life ◽  
A Priori ◽  
Psychological Research ◽  
Attribute Selection ◽  
Specific Class ◽  
Data Mining Techniques ◽  
Using Data

Psychological research centers help indirectly contact professionals from the fields of human life, job environment, family life, and psychological infrastructure for psychiatric patients. This research aims to detect job apathy patterns from the behavior of employee groups in the University of Baghdad and the Iraqi Ministry of Higher Education and Scientific Research. This investigation presents an approach using data mining techniques to acquire new knowledge and differs from statistical studies in terms of supporting the researchers’ evolving needs. These techniques manipulate redundant or irrelevant attributes to discover interesting patterns. The principal issue identifies several important and affective questions taken from a questionnaire, and the psychiatric researchers recommend these questions. Useless questions are pruned using the attribute selection method. Moreover, pieces of information gained through these questions are measured according to a specific class and ranked accordingly. Association and a priori algorithms are used to detect the most influential and interrelated questions in the questionnaire. Consequently, the decisive parameters that may lead to job apathy are determined.

Sign in / Sign up

Export Citation Format

Share Document