Multimedia ontology learning for automatic annotation and video browsing
30 Oct 2008-pp 387-394
TL;DR: This work uses MOWL, a multimedia extension of Web Ontology Language (OWL) which is capable of describing domain concepts in terms of their media properties and of capturing the inherent uncertainties involved.
Abstract: In this work, we offer an approach to combine standard multimedia analysis techniques with knowledge drawn from conceptual metadata provided by domain experts of a specialized scholarly domain, to learn a domain-specific multimedia ontology from a set of annotated examples. A standard Bayesian network learning algorithm that learns structure and parameters of a Bayesian network is extended to include media observables in the learning. An expert group provides domain knowledge to construct a basic ontology of the domain as well as to annotate a set of training videos. These annotations help derive the associations between high-level semantic concepts of the domain and low-level MPEG-7 based features representing audio-visual content of the videos. We construct a more robust and refined version of this ontology by learning from this set of conceptually annotated videos. To encode this knowledge, we use MOWL, a multimedia extension of Web Ontology Language (OWL) which is capable of describing domain concepts in terms of their media properties and of capturing the inherent uncertainties involved. We use the ontology specified knowledge for recognizing concepts relevant to a video to annotate fresh addition to the video database with relevant concepts in the ontology. These conceptual annotations are used to create hyperlinks in the video collection, to provide an effective video browsing interface to the user.
Citations
More filters
Dissertation•
04 Jul 2016
TL;DR: Notre RSD assure un partage de donnees selectif ou les utilisateurs recoivent des contenus selon leurs interets exprimes grâce au paradigme publier/souscrire, en termes de scalabilite, temps de reponse and trafic sur le reseau.
Abstract: Avec l’evolution des nouvelles technologies, les reseaux sociaux (RS) sont frequemment utilises pour partager des donnees entre plusieurs utilisateurs. Ces RS presentent deux limitations majeures. D’une part, une grande partie des contenus partages n’interesse pas ces utilisateurs, ce qui augmente inutilement la consommation de la bande passante et degrade la qualite de service de ces RS. D’autre part, ces RS utilisent une infrastructure centralisee geree par un fournisseur de services. Ce dernier impose des contraintes liees a l’espace de stockage et aux services offerts comme il peut exiger des frais pour l’utilisation du RS. Dans ce contexte, nous avons developpe un RS distribue (RSD) en pair-a-pair (DHT) fonde sur un systeme publier/souscrire. Son service d’evenements est deploye sur les equipements des utilisateurs en les communicant en P2P pour surmonter les contraintes des fournisseurs de services. Notre RSD assure un partage de donnees selectif ou les utilisateurs recoivent des contenus selon leurs interets exprimes grâce au paradigme publier/souscrire. Notre RSD permet de decrire les interets des utilisateurs par des souscriptions semantiques et composites. Pour le traitement de la semantique, nous avons utilise une ontologie de domaine structuree et partagee entre les utilisateurs. Nous avons propose ensuite une methode d’indexation de cette ontologie fondee sur les nombres premiers. Cette methode offre un routage semantique sur une architecture P2P structuree (DHT) de notre plateforme. Pour exprimer les interets composes, nous avons defini une structure de cube pour l’indexation des souscriptions composites sur les noeuds DHT. Cette structure assure le filtrage des evenements composites par une simple recherche binaire dans ce cube. Bien que l’architecture de notre RSD assure l’auto-organisation des noeuds, la disponibilite des donnees et leur livraison sans perte reste encore un defi pour des noeuds tres dynamiques. Nous avons alors propose une strategie de replication selon la disponibilite des noeuds en cherchant a atteindre la disponibilite souhaitee par l’utilisateur. Des experimentations menees a l’aide du simulateur FreePastry ont montre l’efficacite de notre RSD par rapport aux RS existants, en termes de scalabilite, temps de reponse et trafic sur le reseau.
1 citations
Cites background from "Multimedia ontology learning for au..."
...Par exemple, si nous nous intéressons au domaine multimédia (qui couvre déjà les contenus échangés par la majorité des RS), nous pouvons définir notre ontologie de domaine à partir d’une ontologie de domaine définie dans la littérature telle que Multimedia Web Ontology [71]....
[...]
Proceedings Article•
01 Jan 2014TL;DR: In this article, the authors propose the use of Multi-entity Bayesian Networks (MEBNs) for modeling the knowledge and analyzing the content pertaining to the domain of Intangible Cultural Heritage (ICH).
Abstract: In this paper, we propose the use of Multi-entity Bayesian networks (MEBNs) for modeling the knowledge and analyzing the content pertaining to the domain of Intangible Cultural Heritage (ICH). MEBNs provide a rigorous knowledge representation framework in conjunction with reasoning and probabilistic inference capabilities. There are mainly two reasons motivating the use of MEBNs in the domain of ICH. The first is that MEBNs extend first-order logic with the ability to model uncertainty. The second reason is the capability of MEBN to adapt to specific situations by providing custom, situation specific Bayesian networks. Finally, we use an example to demonstrate the potential efficiency of MEBNs in the domain of ICH.
1 citations
09 Dec 2022
TL;DR: In this article , an annotation and retrieval application named NrityaManch is presented, which is dedicated explicitly to the Indian classical dance, focusing on dancer details, dance details, and elements of static dance posture during the annotation.
Abstract: This paper presents an annotation and retrieval application named NrityaManch dedicated explicitly to the Indian classical dance. We primarily choose Bharatanatyam dance for the application development. We exploit ontology technique which captures dance image’s annotation details and structurally organizes the dance database. An OWL2 ontology is developed in Protégé 5.5.0 which is validated using HermiT 1.4.3.456 reasoner to maintain consistency. A user interface is provided for the manual annotation of dance images. Initially, we focus on dancer details, dance details, and elements of static dance posture like hasta mudra during the annotation. All annotation details are saved in RDF/XML file. A search window is provided, which facilitates two types of search - natural language query search and tight query search. Named Entity Recognition (NER) pipeline mechanism is utilized in this work which facilitates keyword extraction from natural language queries. A SPARQL query is automatically generated by the system which is applied to the RDF corpus in order to retrieve distinct images. The NER pipeline mechanism achieves an accuracy of 80% for our dance dataset. The system achieves an average f-score value of 0.8547 for the retrieval functionality. The proposed system intends to help dance learners to find dance resources in a dedicated place and will also help in Indian classical dance preservation.
Patent•
05 May 2014TL;DR: In this article, the authors describe methods and systems effective to monitor a data access activity using ontology-based data access monitoring techniques, such as querying a set of concepts in an ontology, and comparing the planned path with the browsing path.
Abstract: Technologies are generally described for methods and systems effective to monitor a data access activity. In some examples, a method may include receiving, by a processor, a destination concept. The processor may identify a set of concepts, which may include the destination concept and at least one related concept associated with the destination concept, in an ontology. The processor may generate a planned path, which may define a first data access order associated with access of at least one of the related concepts and the destination concept, using the set of concepts. The processor may generate a browsing path which may define a second data access order associated with the data access activity. The processor may compare the planned path with the browsing path. The processor may determine a deviation based on the comparison of the planned path and the browsing path. The processor may monitor the data access activity using the deviation.
References
More filters
01 Aug 1996
TL;DR: A novel addition to the known methods for learning Bayesian networks from data that improves the quality of the learned networks and indicates that learning curves characterizing the procedure that exploits the local structure converge faster than these of the standard procedure.
Abstract: In this paper we examine a novel addition to the known methods for learning Bayesian networks from data that improves the quality of the learned networks. Our approach explicitly represents and learns the local structure in the conditional probability tables (CPTs), that quantify these networks. This increases the space of possible models, enabling the representation of CPTs with a variable number of parameters that depends on the learned local structures. The resulting learning procedure is capable of inducing models that better emulate the real complexity of the interactions present in the data. We describe the theoretical foundations and practical aspects of learning local structures, as well as an empirical evaluation of the proposed method. This evaluation indicates that learning curves characterizing the procedure that exploits the local structure converge faster than these of the standard procedure. Our results also show that networks learned with local structure tend to be more complex (in terms of arcs), yet require less parameters.
544 citations
TL;DR: A method for describing human activities from video images based on concept hierarchies of actions based on semantic primitives, which demonstrates the performance of the proposed method by several experiments.
Abstract: We propose a method for describing human activities from video images based on concept hierarchies of actions. Major difficulty in transforming video images into textual descriptions is how to bridge a semantic gap between them, which is also known as inverse Hollywood problem. In general, the concepts of events or actions of human can be classified by semantic primitives. By associating these concepts with the semantic features extracted from video images, appropriate syntactic components such as verbs, objects, etc. are determined and then translated into natural language sentences. We also demonstrate the performance of the proposed method by several experiments.
364 citations
TL;DR: The OntoLearn system is an infrastructure for automated ontology learning from domain text that uses natural language processing and machine learning techniques, and is part of a more general ontology engineering architecture.
Abstract: Our OntoLearn system is an infrastructure for automated ontology learning from domain text. It is the only system, as far as we know, that uses natural language processing and machine learning techniques, and is part of a more general ontology engineering architecture. We describe the system and an experiment in which we used a machine-learned tourism ontology to automatically translate multiword terms from English to Italian. The method can apply to other domains without manual adaptation.
357 citations
05 Jan 2004
TL;DR: This work proposes to incorporate Bayesian networks (BN), a widely used graphic model for knowledge representation under uncertainty and OWL, the de facto industry standard ontology language recommended by W3C to support uncertain ontology representation and ontology reasoning and mapping.
Abstract: To support uncertain ontology representation and ontology reasoning and mapping, we propose to incorporate Bayesian networks (BN), a widely used graphic model for knowledge representation under uncertainty and OWL, the de facto industry standard ontology language recommended by W3C. First, OWL is augmented to allow additional probabilistic markups, so probabilities can be attached with individual concepts and properties in an OWL ontology. Secondly, a set of translation rules is defined to convert this probabilistically annotated OWL ontology into the directed acyclic graph (DAG) of a BN. Finally, the BN is completed by constructing conditional probability tables (CPT) for each node in the DAG. Our probabilistic extension to OWL is consistent with OWL semantics, and the translated BN is associated with a joint probability distribution over the application domain. General Bayesian network inference procedures (e.g., belief propagation or junction tree) can be used to compute P(C/spl bsol/e): the degree of the overlap or inclusion between a concept C and a concept represented by a description e. We also provide a similarity measure that can be used to find the most similar concept that a given description belongs to.
262 citations
TL;DR: A new learning-oriented model for ontology development and a framework for ontological learning are proposed and important dimensions for classifying ontology learning approaches and techniques are identified.
Abstract: Ontology is one of the fundamental cornerstones of the semantic Web The pervasive use of ontologies in information sharing and knowledge management calls for efficient and effective approaches to ontology development Ontology learning, which seeks to discover ontological knowledge from various forms of data automatically or semi-automatically, can overcome the bottleneck of ontology acquisition in ontology development Despite the significant progress in ontology learning research over the past decade, there remain a number of open problems in this field This paper provides a comprehensive review and discussion of major issues, challenges, and opportunities in ontology learning We propose a new learning-oriented model for ontology development and a framework for ontology learning Moreover, we identify and discuss important dimensions for classifying ontology learning approaches and techniques In light of the impact of domain on choosing ontology learning approaches, we summarize domain characteristics that can facilitate future ontology learning effort The paper offers a road map and a variety of insights about this fast-growing field
211 citations