Learning Discriminative Projections for Text Similarity Measures.
CoNLL '11: Proceedings of the Fifteenth Conference on Computational Natural Language Learning(2011)
摘要
Traditional text similarity measures consider each term similar only to itself and do not model semantic relatedness of terms. We propose a novel discriminative training method that projects the raw term vectors into a common, low-dimensional vector space. Our approach operates by finding the optimal matrix to minimize the loss of the pre-selected similarity function (e.g., cosine) of the projected vectors, and is able to efficiently handle a large number of training examples in the high-dimensional space. Evaluated on two very different tasks, cross-lingual document retrieval and ad relevance measure, our method not only outperforms existing state-of-the-art approaches, but also achieves high accuracy at low dimensions and is thus more efficient.
更多查看译文
关键词
high-dimensional space,low-dimensional vector space,novel discriminative training method,pre-selected similarity function,raw term vector,traditional text similarity measure,training example,ad relevance measure,cross-lingual document retrieval,different task,discriminative projection
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络