Subject estimation system for estimating subject of dialog

A computer automated subject estimation method in a convolution neural network comprising:

• performing, on an input of a word-string vector sequence corresponding to dialog text transcribed from a dialog, topic-dependent convolution process including a convolution operation dependent on a topic by using a processor;
• performing, on the input, topic-independent convolution process including a convolution operation not dependent on the topic by using the processor;
• performing pooling process on outputs of (i) the topic-dependent convolution process and (ii) the topic-independent convolution process by using processor;
• and
• performing full connection process on outputs of the pooling process and estimating a subject label of the dialog by using the processor.

A non-transitory recording medium having a computer program stored thereon, the computer program causing a processor to execute a convolution network for computer automated subject estimation, the convolution network comprising:

• performing, on an input of a word-string vector sequence corresponding to dialog text transcribed from a dialog, a topic-dependent convolution process including a convolution operation dependent on a topic;
• performing, on the input, a topic-independent convolution process including a convolution operation not dependent on the topic;
• performing a process on outputs of (i) the topic-dependent convolution process and (ii) the topic-independent convolution process;
• and
• performing a full connection process on outputs of the pooling process and estimating a subject label of the dialog.