In order to improve the efficiency of ontology construction from heterogeneous knowledge sources, a semantic-based approach is presented. The ontology will be constructed with the application of cluster technique in a...In order to improve the efficiency of ontology construction from heterogeneous knowledge sources, a semantic-based approach is presented. The ontology will be constructed with the application of cluster technique in an incremental way. Firstly, terms will be extracted from knowledge sources and congregate a term set after pretreat-ment. Then the concept set will be built via semantic-based clustering according to semanteme of terms provided by WordNet. Next, a concept tree is constructed in terms of mapping rules between semant^me relationships and concept relationships. The semi-automatic approach can avoid non-consistence due to knowledge engineers having different understanding of the same concept and the obtained ontology is easily to be expanded.展开更多
Existing clothes retrieval methods mostly adopt binary supervision in metric learning.For each iteration,only the clothes belonging to the same instance are positive samples,and all other clothes are“indistinguishabl...Existing clothes retrieval methods mostly adopt binary supervision in metric learning.For each iteration,only the clothes belonging to the same instance are positive samples,and all other clothes are“indistinguishable”negative samples,which causes the following problem.The relevance between the query and candidates is only treated as relevant or irrelevant,which makes the model difficult to learn the continu-ous semantic similarities between clothes.Clothes that do not belong to the same instance are completely considered irrelevant and are uni-formly pushed away from the query by an equal margin in the embedding space,which is not consistent with the ideal retrieval results.Moti-vated by this,we propose a novel method called semantic-based clothes retrieval(SCR).In SCR,we measure the semantic similarities be-tween clothes and design a new adaptive loss based on these similarities.The margin in the proposed adaptive loss can vary with different se-mantic similarities between the anchor and negative samples.In this way,more coherent embedding space can be learned,where candidates with higher semantic similarities are mapped closer to the query than those with lower ones.We use Recall@K and normalized Discounted Cu-mulative Gain(nDCG)as evaluation metrics to conduct experiments on the DeepFashion dataset and have achieved better performance.展开更多
Multimedia document annotation is used in traditional multimedia databasesystems. However, without the help of human beings, it is very difficult to extract the semanticcontent of multimedia automatically. On the othe...Multimedia document annotation is used in traditional multimedia databasesystems. However, without the help of human beings, it is very difficult to extract the semanticcontent of multimedia automatically. On the other hand, it is a tedious job to annotate multimediadocuments in large databases one by one manually. This paper first introduces a method to constructa semantic net-work on top of a multimedia database. Second, a useful and efficient annotationstrategy is presented based on the framework to obtain an accurate and rapid annotation of anymultimedia databases. Third, two methods of joint similarity measures for semantic and low-levelfeatures are evaluated .展开更多
文摘In order to improve the efficiency of ontology construction from heterogeneous knowledge sources, a semantic-based approach is presented. The ontology will be constructed with the application of cluster technique in an incremental way. Firstly, terms will be extracted from knowledge sources and congregate a term set after pretreat-ment. Then the concept set will be built via semantic-based clustering according to semanteme of terms provided by WordNet. Next, a concept tree is constructed in terms of mapping rules between semant^me relationships and concept relationships. The semi-automatic approach can avoid non-consistence due to knowledge engineers having different understanding of the same concept and the obtained ontology is easily to be expanded.
文摘Existing clothes retrieval methods mostly adopt binary supervision in metric learning.For each iteration,only the clothes belonging to the same instance are positive samples,and all other clothes are“indistinguishable”negative samples,which causes the following problem.The relevance between the query and candidates is only treated as relevant or irrelevant,which makes the model difficult to learn the continu-ous semantic similarities between clothes.Clothes that do not belong to the same instance are completely considered irrelevant and are uni-formly pushed away from the query by an equal margin in the embedding space,which is not consistent with the ideal retrieval results.Moti-vated by this,we propose a novel method called semantic-based clothes retrieval(SCR).In SCR,we measure the semantic similarities be-tween clothes and design a new adaptive loss based on these similarities.The margin in the proposed adaptive loss can vary with different se-mantic similarities between the anchor and negative samples.In this way,more coherent embedding space can be learned,where candidates with higher semantic similarities are mapped closer to the query than those with lower ones.We use Recall@K and normalized Discounted Cu-mulative Gain(nDCG)as evaluation metrics to conduct experiments on the DeepFashion dataset and have achieved better performance.
文摘Multimedia document annotation is used in traditional multimedia databasesystems. However, without the help of human beings, it is very difficult to extract the semanticcontent of multimedia automatically. On the other hand, it is a tedious job to annotate multimediadocuments in large databases one by one manually. This paper first introduces a method to constructa semantic net-work on top of a multimedia database. Second, a useful and efficient annotationstrategy is presented based on the framework to obtain an accurate and rapid annotation of anymultimedia databases. Third, two methods of joint similarity measures for semantic and low-levelfeatures are evaluated .
