scholarly journals mlCAF: Multi-Level Cross-Domain Semantic Context Fusioning for Behavior Identification

Sensors ◽  
2017 ◽  
Vol 17 (10) ◽  
pp. 2433 ◽  
Author(s):  
Muhammad Razzaq ◽  
Claudia Villalonga ◽  
Sungyoung Lee ◽  
Usman Akhtar ◽  
Maqbool Ali ◽  
...  
Keyword(s):  
Semantic Context ◽  
Cross Domain ◽  
Multi Level

scholarly journals CONTRASTING ENGLISH AND SERBIAN THROUGH THE LENSES OF MULTILEVEL-GROUNDED SEMANTICS

2021 ◽  
Vol 18 (48) ◽  
pp. 167-183
Author(s):  
Mihailo Antović ◽  
Keyword(s):  
Mental Representations ◽  
General Knowledge ◽  
Embodied Experience ◽  
Modal Interaction ◽  
Implicit Information ◽  
Two Agents ◽  
Cross Domain ◽  
Grammatical Construction ◽  
Multi Level ◽  
Grounded Semantics

The paper illustrates how the author’s emerging theory of “multi-level grounding” may be applied to some contrastive phenomena in English and Serbian. The theory argues that classic semantic approaches based on cross-space interaction may profit enormously from a more thorough consideration of contextual constraints on meaning generation. For example, to understand even a fairly simple comparison such as “Achilles is a lion”, one needs to know a lot more than just how to, depending on the paradigm of choice, “cross-domain map”, “blend”, or “analogize” appropriate formal elements of the two concepts understood as mere mental representations. Rather, to be meaningful in more than just an academic sense, the interpretation needs to call layers of context, from the very general knowledge of who Achilles is and what lions are to specific cultural and even personal connotations appropriate to the two agents and their interaction. In relation to the earlier work of Searle and Langacker, cognitive linguists Coulson and Oakley propose to allocate such knowledge to the construct of the “grounding box” (containing implicit information on the agents, forum, and circumstances surrounding the utterance). The author’s theory makes this concept more refined, suggesting a series of at least six hierarchical and partly recursive grounding boxes constraining meaning generation – from the perceptual attributes of objects cognized to such percepts’ cross-modal interaction with the interlocutors’ embodied experience, to their affective, conceptual, and discourse-driven (re) interpretations. The analysis in this paper aims to show how this approach may be instrumental in disentangling the (seemingly) shared and different semantic strategies in the way English and Serbian treat a simple stock expression (“You are right” / “U pravu si”), grammatical construction (“tolerant of” / “tolerantan prema”), and widely used idiom (“a finger in every pie” / “u svakoj čorbi mirođija”).

scholarly journals Image Caption Generation Using Multi-Level Semantic Context Information

Symmetry ◽  
2021 ◽  
Vol 13 (7) ◽  
pp. 1184
Author(s):  
Peng Tian ◽  
Hongwei Mo ◽  
Laihao Jiang
Keyword(s):  
Object Detection ◽  
Image Features ◽  
Semantic Context ◽  
Context Information ◽  
Semantic Features ◽  
Image Captioning ◽  
Scene Image ◽  
Visual Tasks ◽  
Multi Level ◽  
Highly Correlated

Object detection, visual relationship detection, and image captioning, which are the three main visual tasks in scene understanding, are highly correlated and correspond to different semantic levels of scene image. However, the existing captioning methods convert the extracted image features into description text, and the obtained results are not satisfactory. In this work, we propose a Multi-level Semantic Context Information (MSCI) network with an overall symmetrical structure to leverage the mutual connections across the three different semantic layers and extract the context information between them, to solve jointly the three vision tasks for achieving the accurate and comprehensive description of the scene image. The model uses a feature refining structure to mutual connections and iteratively updates the different semantic features of the image. Then a context information extraction network is used to extract the context information between the three different semantic layers, and an attention mechanism is introduced to improve the accuracy of image captioning while using the context information between the different semantic layers to improve the accuracy of object detection and relationship detection. Experiments on the VRD and COCO datasets demonstrate that our proposed model can leverage the context information between semantic layers to improve the accuracy of those visual tasks generation.

Sign in / Sign up

Export Citation Format

Share Document