基于特征融合的社区问答问句相似度计算

Question Similarity Calculation Based on Feature Fusion in Community Question Answering

提出了一种基于特征融合的问句匹配框架来解决问句相似度检测方法,利用答案特征、词序特征、统计特征和语义特征相结合来解决问句相似度计算问题.在Yahoo!Answers上抽取的真实标注数据集上进行实验,实验结果表明:该方法在性能上得到了较好的结果.

A testing method based on the framework combining different features with the characteristic of the answer,the word order characteristic,the statistical characteristic and semantic characteristic is proposed to caculate question simularity of question answering.The experiments on the true labeling dataset extracted from the Yahoo! Answers shows that the proposed method achieves a better performance.