To realize edit processing, in which a user can freely edit solution data used by summary processing with a machine learning method, and summary processing dedicated to the user using the above solution data.

An evaluation customizing means 110 extracts sentence data in a range designated by the user from a text/summary 4, as the user designated summary, displays an input item for evaluation for each of multiple summary properties, receives the evaluation input by the user regarding the user designated summary, and generates the summary for the text by the automatic summary generation processing by a designated rule or the machine learning method. The solution data including data, assigned with the solution which is the user input evaluation to the problem comprising the text and the user designated summary, and data, assigned with the solution which is the bad evaluation indicating not the user designated summary to the problem comprising the text and portions other than the user designated summary are generated, and outputted to a solution data storage part 120.