Document search device, document search system, document search program, and document search method

A document search system comprising:

• a document search device;
• and
• a user terminal,
• wherein the document search device includes:
• a memory configured to store a program, and
• a processor configured to execute the program stored in the memory and perform a process including
• receiving a keyword of a document search, the keyword consisting of text characters that are an arbitrary combination of letters, numbers, and symbols;
• acquiring, from a text document, a hit character string matching a character string having a portion of the text characters of the keyword replaced with a wildcard which is a special text character considered to match all text characters during the document search, and character strings of the text document before and after the hit character string;
• computing a forward likelihood of the hit character string based on the character string before the hit character string;
• computing a reverse likelihood of the hit character string based on a reversed character string after the hit character string, obtained by rearranging text characters of the character string after the hit character string, so that an order of the text characters of the character string after the hit character string is reversed and the text character at an end of the character string becomes the text character at a beginning of the reversed character string and the text character at a beginning of the character string becomes the text character at an end of the reversed character string;
• computing a likelihood of the hit character string based on both the forward likelihood computed by the computing the forward likelihood and the reverse likelihood computed by the computing the reverse likelihood;
• and
• outputting a result of the document search based on the likelihood of the hit character string computed by the computing the likelihood based on both the forward likelihood and the reverse likelihood, wherein:
• the computing the forward likelihood computes the forward likelihood for a case where the hit character string is assumed to be a search keyword, based on the character string before the hit character string by computing a likelihood of a beginning text character of the hit character string, adding the beginning text character of the hit character string to an end of the input character string before the hit character string and excluding a beginning text character of the input character string before the hit character string to compute a likelihood of a second text character counted from the beginning of the hit character string, and performing a similar process with respect to all of a third and subsequent text characters counted from the beginning of the hit character string, and
• the computing the reverse likelihood computes the reverse likelihood for the case where the hit character string is assumed to be the search keyword, based on the reversed character string after the hit character string by computing a likelihood of an end text character of the hit character string, adding the end text character of the hit character string to an end of the reversed character string after the hit character string and excluding a beginning text character of the reversed character string after the hit character string to compute a likelihood of a second text character counted from the end of the hit character string, and performing a similar process with respect to all of a third and subsequent text characters counted from the end of the hit character string.

A non-transitory computer-readable recording medium storing a program which, when executed by a computer, causes the computer to perform a process comprising:

• receiving a keyword of a document search, the keyword consisting of text characters that are an arbitrary combination of letters, numbers, and symbols;
• acquiring, from a text document, a hit character string matching a character string having a portion of the text characters of the keyword replaced with a wildcard which is a special text character considered to match all text characters during the document search, and character strings of the text document before and after the hit character string;
• computing a forward likelihood of the hit character string based on the character string before the hit character string;
• computing a reverse likelihood of the hit character string based on a reversed character string after the hit character string, obtained by rearranging text characters of the character string after the hit character string, so that an order of the text characters of the character string after the hit character string is reversed and the text character at an end of the character string becomes the text character at a beginning of the reversed character string and the text character at a beginning of the character string becomes the text character at an end of the reversed character string;
• computing a likelihood of the hit character string based on both the forward likelihood computed by the computing the forward likelihood and the reverse likelihood computed by the computing the reverse likelihood;
• and
• outputting a result of the document search based on the likelihood of the hit character string computed by the computing the likelihood based on both the forward likelihood and the reverse likelihood, wherein:
• the computing the forward likelihood computes the forward likelihood for a case where the hit character string is assumed to be a search keyword, based on the character string before the hit character string by computing a likelihood of a beginning text character of the hit character string, adding the beginning text character of the hit character string to an end of the input character string before the hit character string and excluding a beginning text character of the input character string before the hit character string to compute a likelihood of a second text character counted from the beginning of the hit character string, and performing a similar process with respect to all of a third and subsequent text characters counted from the beginning of the hit character string, and
• the computing the reverse likelihood computes the reverse likelihood for the case where the hit character string is assumed to be the search keyword, based on the reversed character string after the hit character string by computing a likelihood of an end text character of the hit character string, adding the end text character of the hit character string to an end of the reversed character string after the hit character string and excluding a beginning text character of the reversed character string after the hit character string to compute a likelihood of a second text character counted from the end of the hit character string, and performing a similar process with respect to all of a third and subsequent text characters counted from the end of the hit character string.

A computer-implemented method for performing a document search, the computer-implemented method comprising:

• receiving, by the computer, a keyword of a document search, the keyword consisting of text characters that are an arbitrary combination of letters, numbers, and symbols;
• acquiring, by the computer from a text document, a hit character string matching a character string having a portion of the text characters of the keyword replaced with a wildcard which is a special text character considered to match all text characters during the document search, and character strings of the text document before and after the hit character string;
• computing, by the computer, a forward likelihood of the hit character string based on the character string before the hit character string;
• computing, by the computer, a reverse likelihood of the hit character string based on a reversed character string after the hit character string, obtained by rearranging text characters of the character string after the hit character string, so that an order of the text characters of the character string after the hit character string is reversed and the text character at an end of the character string becomes the text character at a beginning of the reversed character string and the text character at a beginning of the character string becomes the text character at an end of the reversed character string;
• computing, by the computer, a likelihood of the hit character string based on both the forward likelihood computed by the computing the forward likelihood and the reverse likelihood computed by the computing the reverse likelihood;
• and
• outputting, by the computer, a result of the document search based on the likelihood of the hit character string computed by the computing the likelihood based on both the forward likelihood and the reverse likelihood, wherein:
• the computing the forward likelihood computes the forward likelihood for a case where the hit character string is assumed to be a search keyword, based on the character string before the hit character string by computing a likelihood of a beginning text character of the hit character string, adding the beginning text character of the hit character string to an end of the input character string before the hit character string and excluding a beginning text character of the input character string before the hit character string to compute a likelihood of a second text character counted from the beginning of the hit character string, and performing a similar process with respect to all of a third and subsequent text characters counted from the beginning of the hit character string, and
• the computing the reverse likelihood computes the reverse likelihood for the case where the hit character string is assumed to be the search keyword, based on the reversed character string after the hit character string by computing a likelihood of an end text character of the hit character string, adding the end text character of the hit character string to an end of the reversed character string after the hit character string and excluding a beginning text character of the reversed character string after the hit character string to compute a likelihood of a second text character counted from the end of the hit character string, and performing a similar process with respect to all of a third and subsequent text characters counted from the end of the hit character string.