WikiKreator: Improving Wikipedia Stubs Automatically

Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing, pages 867–877, Beijing, China, July 26-31, 2015. c⃝2015 Association for Computational Linguistics WikiKreator: Improving Wikipedia Stubs Automatically Siddhartha Banerjee The Pennsylvania State University Information Sciences and Technology University Park, PA, USA sub253@ist.psu.edu Prasenjit Mitra Qatar Computing Research Institute Hamad Bin Khalifa University Doha, Qatar pmitra@qf.org.qa Abstract Stubs on Wikipedia often lack comprehen- sive information. The huge cost of edit- ing Wikipedia and the presence of only a limited number of active contributors curb the consistent growth of Wikipedia. In this work, we present WikiKreator, a system that is capable of generating content au- tomatically to improve existing stubs on Wikipedia. The system has two compo- nents. First, a text classifier built using topic distribution vectors is used to as- sign content from the web to various sec- tions on a Wikipedia article. Second, we propose a novel abstractive summariza- tion technique based on an optimization framework that generates section-specific summaries for Wikipedia stubs. Experi- ments show that WikiKreator is capable of generating well-formed informative con- tent. Further, automatically generated con- tent from our system have been appended to Wikipedia stubs and the content has been retained successfully proving the ef- fectiveness of our approach. 1 Introduction Wikipedia provides comprehensive information on various topics. However, a significant percent- age of the articles are stubs1 that require exten- sive effort in terms of adding and editing content to transform them into complete articles. Ideally, we would like to create an automatic Wikipedia content generator, which can generate a compre- hensive overview on any topic using available in- formation from the web and append the gener- ated content to the stubs. Addition of automati- cally generated content can provide a useful start- 1https://en.wikipedia.org/wiki/ Wikipedia:Stub ing point for contributors on Wikipedia, which can be improved upon later. Several approaches to automatically generate Wikipedia articles have been explored (Sauper and Barzilay, 2009; Banerjee et al., 2014; Yao et al., 2011). To the best of our knowledge, all the above mentioned methods identify informa- tion sources from the web using keywords and directly use the most relevant excerpts in the fi- nal article. Information from the web cannot be directly copied into Wikipedia due to copy- right violation issues (Banerjee et al., 2014). Further, keyword search does not always sat- isfy information requirements (Baeza-Yates et al., 1999). To address the above-mentioned issues, we present WikiKreator – a system that can au- tomatically generate content for Wikipedia stubs. First, WikiKreator does not operate using keyword search. Instead, we use a classifier trained using topic distribution features to identify relevant con- tent for the stub. Topic-distribution features are more effective than keyword search as they can identify relevant content based on word distribu- tions (Song et al., 2010). Second, we propose a novel abstractive summarization (Dalal and Malik, 2013) technique to summarize content from mul- tiple snippets of relevant information.2 Figure 1 shows a stub that …

Content truncated to 500 words