Extraction of reported speeches from Arabic Lebanese newspapers

This paper presents ongoing work on the extraction of Arabic reported speech, made by Lebanese politicians, from Arabic Lebanese newspapers. This work is part of a functional system for extraction, presentation and archiving of reported speech made by Lebanese politicians, which constitutes a valuable resource for political analysts, press agents, company researchers and political actors. The system automatically identifies about 280 reported speeches per day from about 1,000 newspaper articles, together with their referents, all are correctly identified as reported speech but only about 200 are correctly referred to their referents. The correctly identified reported speeches that refer to the correctly identified referents are then submitted to a web-based application, which is publicly accessible at http://citations-explorer.com/lpc/.