BEGIN:VCALENDAR
PRODID:-//hacksw/handcal//NONSGML v1.0//EN
METHOD:PUBLISH
BEGIN:VEVENT
DESCRIPTION:Click for Latest Location Information: http://smartdata2017.dataversity.net/sessionPop.cfm?confid=110&proposalid=9370\nThe detection of table lines in an unstructured document has two major components. First, during a feature extraction phase, text lines are described by a set of features (derived from the table model), and then a deep neural network classifier is trained to distinguish between table and non-table lines. Training of a classifier is justified by the large diversity of table structures which make very difficult to define clear cuts and conditions to separate tables from non-tables, especially when a large number of descriptors are used to characterize a text line.\nThe TABSTRACT SaaS is used to extract the tables from an unstructured document and store the results in a spreadsheet.  The extractors can be also accessed via an API.\nThe extractor is integrated with the Recognos ETI Data Extraction Platform.
DTSTART:20170131T170000
SUMMARY:TABSTRACT Table Extraction Service From Unstructured Documents
DTEND:20170131T174459
LOCATION: See Description
END:VEVENT
END:VCALENDAR