{"title":"Text classification based on limited bibliographic metadata","authors":"K. Denecke, T. Risse, Thomas Baehr","doi":"10.1109/ICDIM.2009.5356767","DOIUrl":null,"url":null,"abstract":"In this paper, we introduce a method for categorizing digital items according to their topic, only relying on the document's metadata, such as author name and title information. The proposed approach is based on a set of lexical resources constructed for our purposes (e.g., journal titles, conference names) and on a traditional machine-learning classifier that assigns one category to each document based on identified core features. The system is evaluated on a real-world data set and the influence of different feature combinations and settings is studied. Although the available information is limited, the results show that the approach is capable to efficiently classify data items representing documents.","PeriodicalId":300287,"journal":{"name":"2009 Fourth International Conference on Digital Information Management","volume":"1 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2009-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"6","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2009 Fourth International Conference on Digital Information Management","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/ICDIM.2009.5356767","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 6
Abstract
In this paper, we introduce a method for categorizing digital items according to their topic, only relying on the document's metadata, such as author name and title information. The proposed approach is based on a set of lexical resources constructed for our purposes (e.g., journal titles, conference names) and on a traditional machine-learning classifier that assigns one category to each document based on identified core features. The system is evaluated on a real-world data set and the influence of different feature combinations and settings is studied. Although the available information is limited, the results show that the approach is capable to efficiently classify data items representing documents.