Analysis of sentiment on newspaper quotations: A preliminary experiment

Most Opinion Mining work has been carried out on highly subjective text types where the target is clearly defined and unique across the text like movie or product reviews. However, when applying Opinion Mining to the news domain, it is necessary to clearly define the essence of the task, in a more precise manner than it has been done in the field so far. The main tasks that have been implemented in this paper are: identifying a target, creating three different datasets of quotations (reported speech), pertaining to the identified target, from three different dailies, separation of good and bad news content from the good and bad emotions expressed on the target and comparing the results with manual work.