Text mining system for analysis target data, a text mining method for analysis target data and a recording medium for recording analysis target data

A text mining method performed by at least one processor, the method comprising:

• judging whether there is a commonality in expressions among pieces of text data, said pieces of text data being included in plural pieces of analysis target data, respectively;
• generating a plurality of common analysis viewpoints which is a condition to extract an expression from each of said pieces of analysis target data with said commonality
• identifying, for each of said plurality of common analysis viewpoints, one or more positive example sets each including an expression matching a corresponding one of said plurality of common analysis viewpoints in each of said pieces of analysis target data with said commonality;
• calculating, for each of said plurality of common analysis viewpoints, characteristic quantity showing a degree of characteristic expression of said one or more positive example sets for each of expressions in each of said pieces of analysis target data with said commonality;
• providing, for each of said plurality of common analysis viewpoints, ranks for said expressions in descending order of said calculated characteristic quantity;
• and
• extracting a common analysis viewpoint for said pieces of analysis target data with said commonality from said plurality of common analysis viewpoints in case that a difference in ranks provided for each of said expressions is equal to or greater than a predetermined threshold for said common analysis viewpoint.

A non-transitory computer readable storage medium having recorded thereon a program for text mining, the program when executed by a processor causing a computer to perform the text mining, comprising:

• judging whether there is a commonality in expressions among pieces of text data, said pieces of text data being included in plural pieces of analysis target data, respectively;
• generating a plurality of common analysis viewpoints which is a condition to extract an expression from each of said pieces of analysis target data with said commonality;
• identifying, for each of said plurality of common analysis viewpoints, one or more positive example sets each including an expression matching a corresponding one of said plurality of common analysis viewpoints in each of said pieces of analysis target data with said commonality;
• calculating, for each of said plurality of common analysis viewpoints, characteristic quantity showing a degree of characteristic expression of said one or more positive example sets for each of expressions in each of said pieces of analysis target data with said commonality;
• providing, for each of said plurality of common analysis viewpoints, ranks for said expressions in descending order of said calculated characteristic quantity;
• and
• extracting a common analysis viewpoint for said pieces of analysis target data with said commonality from said plurality of common lysis viewpoints in case that a difference in ranks provided for each of said expressions is equal to or greater than a predetermined threshold for said common analysis viewpoint.

A text mining system, the system implemented by at least one processor, the system comprising:

• analysis target data pair search processor configured to judge whether there is a commonality in expressions among pieces of text data, said pieces of text data being included in plural pieces of analysis target data, respectively;
• analysis viewpoint generation processor configured to generate a plurality of common analysis viewpoints which is a condition to extract an expression from each of said pieces of analysis target data with said commonality;
• positive example set identification processor configured to identify, for each of said plurality of common analysis viewpoints, one or more positive example sets each including an expression matching a corresponding one of said plurality of common analysis viewpoints in each of said pieces of analysis target data with said commonality;
• characteristic quantity calculation configured to calculate, for each of said plurality of common analysis viewpoints, characteristic quantity showing a degree of characteristic expression of said one or more positive example sets for each of expressions in each of said pieces of analysis target data with said commonality;
• and
• characteristic expression ranking processor configured to provide, for each of said plurality of common analysis viewpoints, ranks for said expressions in descending order of said calculated characteristic quantity, wherein
• said analysis target data pair search processor is configured to extract a common analysis viewpoint for said pieces of analysis target data with said commonality from said plurality of common analysis viewpoints in case that a difference in ranks provided for each of said expressions is equal to or greater than a predetermined threshold for said common analysis viewpoint.