Dealing with issues such as too simple image features and word noise inference in product image sentence anmotation, a product image sentence annotation model focusing on image feature learning and key words summariza...Dealing with issues such as too simple image features and word noise inference in product image sentence anmotation, a product image sentence annotation model focusing on image feature learning and key words summarization is described. Three kernel descriptors such as gradient, shape, and color are extracted, respectively. Feature late-fusion is executed in turn by the multiple kernel learning model to obtain more discriminant image features. Absolute rank and relative rank of the tag-rank model are used to boost the key words＇ weights. A new word integration algorithm named word sequence blocks building （WSBB） is designed to create N-gram word sequences. Sentences are generated according to the N-gram word sequences and predefined templates. Experimental results show that both the BLEU-1 scores and BLEU-2 scores of the sentences are superior to those of the state-of-art baselines.展开更多
Web 2.0应用的兴起,推进了情报学科由＂文献组织＂向＂知识组织＂演化。网页标签作为重要的Web2.0应用之一,已经成为大众组织知识的常用途径。然而,现有的标签排序方法难以有效满足知识组织的需求。本文在三核协同标签模型的基础上,充...Web 2.0应用的兴起,推进了情报学科由＂文献组织＂向＂知识组织＂演化。网页标签作为重要的Web2.0应用之一,已经成为大众组织知识的常用途径。然而,现有的标签排序方法难以有效满足知识组织的需求。本文在三核协同标签模型的基础上,充分考虑标签和用户、标签和标签、标签和文档之间的关系,提出了一种结合HITS和随机跳转的标签排序方法。该方法利用高质量标签和高质量用户之间的相互加强关系,根据标签之间的相似性来找出高质量相关标签,有效提高标签排序的质量。在Delicious数据集上的实验结果表明,该方法能较大提高标签排序的准确度。展开更多
基金The National Natural Science Foundation of China(No.61133012)the Humanity and Social Science Foundation of the Ministry of Education(No.12YJCZH274)+1 种基金the Humanity and Social Science Foundation of Jiangxi Province(No.XW1502,TQ1503)the Science and Technology Project of Jiangxi Science and Technology Department(No.20121BBG70050,20142BBG70011)
文摘Dealing with issues such as too simple image features and word noise inference in product image sentence anmotation, a product image sentence annotation model focusing on image feature learning and key words summarization is described. Three kernel descriptors such as gradient, shape, and color are extracted, respectively. Feature late-fusion is executed in turn by the multiple kernel learning model to obtain more discriminant image features. Absolute rank and relative rank of the tag-rank model are used to boost the key words＇ weights. A new word integration algorithm named word sequence blocks building （WSBB） is designed to create N-gram word sequences. Sentences are generated according to the N-gram word sequences and predefined templates. Experimental results show that both the BLEU-1 scores and BLEU-2 scores of the sentences are superior to those of the state-of-art baselines.
