[Patent Document 2] PCT International Publication No. WO2014/109127

[Patent Document 3] Japanese Patent Application Publication No. 2015-49574

The retrieval apparatus described in Patent Document 1 retrieves moving image data by using the feature value of a person specified by the user as a key. In the case of using this retrieval apparatus, the user needs to specify the person to be retrieved for in advance and provide the retrieval apparatus with the feature value of the person. In a case where a person to be retrieved is not identified, the retrieval apparatus cannot be used.

A similar problem occurs in the case of analyzing text data. That is, in a case where a word or a sentence to be retrieved is specified in advance, it is possible to retrieve the desired word or sentence from the text data by providing the retrieval apparatus with the word or the sentence as a retrieval key. However, in a case where a word or a sentence to be retrieved is not identified in advance, the retrieval cannot be executed.

In addition, a similar problem occurs in the case of analyzing other pieces of data, such as voice data, music data, image data, figure data, fingerprint data, biometric information, time series data (stock price fluctuation time series data and the like), file archive, object file, and binary data.

It is an object of the present invention to provide a new technique for retrieving a subject satisfying a predetermined condition from data.