@conference {6362, title = {Data Mining Historical Newspaper Metadata - Old News Teaches History}, booktitle = {IFLA News Media Section Conference }, year = {2016}, month = {2016/04/21}, publisher = {Staats- und Universit{\"a}tsbibliothek Hamburg Carl von Ossietzky }, organization = {Staats- und Universit{\"a}tsbibliothek Hamburg Carl von Ossietzky }, address = {Hamburg}, abstract = {

In this age of Big Data this paper describes how the state-of-the-art OLR (optical layout recognition) technique in one of the largest heritage press digitization projects in Europe (www.europeana-newspapers.eu, 2012-2015) was used in a data mining experiment. Data analysis was applied to descriptive metadata (number of pages, articles, words, illustrations, ads{\textellipsis}) derived from a subset of the Europeana Newspapers collection. The METS/ALTO XML data from a 850K page subset of six XIXth-XXth century French newspaper titles from the collection was analyzed with data mining and data visualization techniques that show promising ways for the production of knowledge about historical newspapers that are of great interest for digital libraries (digitization programs management, curation, and mediation of newspaper collections) as well as for the digital humanities. Equipped with basic tools widely used in libraries (XSL, spreadsheet, charts generator), we show that simple newspaper metadata can give insights into the history of the press and into history itself.

}, keywords = {ALTO, data mining, data visualisation, digital libraries, metada, METS, OCR, OLR, press}, url = {http://altomator.github.io/EN-data_mining/}, author = {Jean-Philippe Moreux} }