Hasso-Plattner-Institut
 
    • de
 

Complete List of Conference Papers of the chair of Prof. Dr. Christoph Meinel

Here you can find all our peer-reviewed conference papers:

Does Multilevel Semantic Representation Improve Text Categorization?

Wang, Cheng; Yang, Haojin; Meinel, Christoph in Proceedings of the 26th International Conference on Database and Expert Systems Applications (DEXA 2015) Volume   9261   von   Lecture Notes in Computer Science 9261 , Seite 319-333 . Springer , 2015 .

This paper presents a novel approach for text categorization by fusing “Bag-of-words” (BOW) word feature and multilevel semantic feature (SF). By extending Online LDA (OLDA) as multilevel topic model for learning a semantic space with different topic granularity, multilevel semantic features are extracted for representing text component. The effectiveness of our approach is evaluated on both large scale Wikipedia corpus and middle-sized 20newsgroups dataset. The former experiment shows that our approach is able to preform semantic feature extraction on large scale dataset. It also demonstrates the topics generated from different topic level have different semantic scopes, which is more appropriate to represent text content. Our classification experiments on 20newsgroups achieved 82.19 % accuracy, which illustrates the effectiveness of fusing BOW and SF features. The further investigation on word and semantic feature fusion proves that Support Vector Machine (SVM) is more sensitive to semantic feature than Naive Bayes (NB), K Nearest Neighbor(KNN), Decision Tree (DT). It is shown that appropriately fusing low-level word feature and high-level semantic feature can achieve equally well or even better result than state-of-the-art with reduced feature dimension and computational complexity
Weitere Informationen
Tags its
BibTeX

You can also find sub-lists of our peer-reviewed conference papers focusing on 

Other Links

... to our Research
              Security Engineering - Learning & Knowledge Tech - Design Thinking - former
... to our Teaching
              Tele-Lectures - MOOCs - Labs - Systems 
... to our Publications
              Books - Journals - Conference-Papers - Patents
... and to our Annual Reports.