Text Feature Extraction and Selection Based on Attention Mechanism
Selecting features that represent a particular corpus is important to the success of many machine learning and text mining applications.However,the previous attention-based work only focused on feature augmentation in the lexical level,lacking the exploration of feature enhancement in the sentence level.In this paper,we exploit a novel feature extraction and selection model for information retrieval,denoted by Dynamic Feature Generation Network(DFGN).In sentence dimension,features are firstly extracted by a variety of different attention mechanisms,then dynamically filtered by thresholds automatically learned.Different kinds of characteristics are distilled according to specific tasks,enhancing the practicability and robustness of the model.DFGN relies solely on the text itself,requires no external feature engineering.Our approach outperforms previous work on multiple well-known answer selection datasets.Through the analysis of the experiments,we prove that DFGN provides excellent retrieval and interpretative abilities.
Feature extraction and selection Machine learning Question answering
Longxuan Ma Lei Zhang
Beijing University of Posts and Telecommunications,Beijing,China
国际会议
澳门
英文
615-627
2019-04-14（万方平台首次上网日期，不代表论文的发表时间）