This invention is to automatically extract noteworthy information from a large amount of information. First, a disclosure unit of an individual opinion such as a statement in a personal web page or a bulletin board is extracted from collected contents information, and information (URL and/or statement number) to specify the disclosure unit of the individual opinion is registered. Next, an object (company name and/or industry type) of the individual opinion is specified. Then, the disclosed contents of the individual opinion are analyzed, so that an evaluation as to the object (good evaluation / bad evaluation) is specified. Besides, the reliability is determined based on referenced degree ranking and based on whether information to indicate the basis of the opinion and/or the identity of the speaker is included. Thus, the evaluation as to the object as characteristics of the individual opinion can be presented to requesters. Besides, for example, only a bad evaluation can be extracted from evaluations as to the object of the individual opinion. Furthermore, the opinion, which has a high influence degree and is noteworthy, can also be found based on the referenced degree ranking and/or the reliability.