Recognizing Biographical Sections in Wikipedia

Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, pages 811–816, Lisbon, Portugal, 17-21 September 2015. c⃝2015 Association for Computational Linguistics. Recognizing Biographical Sections in Wikipedia Alessio Palmero Aprosio Fondazione Bruno Kessler Via Sommarive, 18 38123 Trento, Italy aprosio@fbk.eu Sara Tonelli Fondazione Bruno Kessler Via Sommarive, 18 38123 Trento, Italy satonelli@fbk.eu Abstract Wikipedia is the largest collection of ency- clopedic data ever written in the history of humanity. Thanks to its coverage and its availability in machine-readable format, it has become a primary resource for large- scale research in historical and cultural studies. In this work, we focus on the subset of pages describing persons, and we investigate the task of recognizing bi- ographical sections from them: given a person’s page, we identify the list of sec- tions where information about her/his life is present. We model this as a sequence classification problem, and propose a su- pervised setting, in which the training data are acquired automatically. Besides, we show that six simple features extracted only from the section titles are very infor- mative and yield good results well above a strong baseline. 1 Introduction In the last years, several projects have started to address the mechanisms behind cultural de- velopment, borrowing techniques and algorithms from computer science and natural language pro- cessing to serve historical investigation. Efforts such as BiographyNet1, Pantheon2 or the Aus- trian Prosopographical Information System3 prove an increasing interest in automatically extracting biographical descriptions from large amounts of data and combining them in a more general pic- ture, taking advantage of the availability of such descriptions on the web. Wikipedia has been 1http://www.biographynet.nl 2http://pantheon.media.mit.edu/ treemap/country_exports/IQ/all/-4000/ 2010/H15/pantheon 3http://www.oeaw.ac.at/acdh/de/node/ 188 the main source of information for research in this direction despite its many biases, for in- stance its well-known English, Western and gen- der bias (Wikipedia Contributors, 2014). In fact, Wikipedia coverage both in terms of pages and in terms of languages, as well as the structured in- formation that can be leveraged through DBpedia, has made it the primary resource for large-scale analyses on biographies. However, the lack of a consistent template for describing persons’ lives led to the creation of a plethora of page types, where biographical information is displayed in di- verse ways. Based on a random sample of 100 persons’ pages, we noticed that only 20% of them includes a section called Biography or Life, typically con- taining a set of subsections describing the main periods in a person’s life from birth to death (see for instance https://en.wikipedia.org/ wiki/Leonard_Bernstein). The other pages in our sample do not follow a pre-defined pattern and present the person’s biography in one or several sections at the same level of the other ones (see for instance https://en. wikipedia.org/wiki/Judy_Holliday, with the Filmography and Discography sections at the same level of Early Life and Career). Given this high variability, it is very difficult to extract all and only those sections that describe a biography, and that build all together in sequence the description of a person’s life. …

Content truncated to 500 words