Faq consolidation assistance device, faq consolidation assistance method, and program

A computer-implemented method for maintaining frequently asked questions (FAQ) including a pair of a question item and a response, the method comprising:

• extracting from reception log data, customer question items that have been determined as question items posed by a customer, from contents of an inquiry from the customer and a response to the customer, wherein the extracting of the customer question items deletes customer question items that are not associated with the search queries among the extracted customer question items;
• extracting, from click action log data, search queries previously executed to search for question items of FAQ in an FAQ search system, wherein the search queries are based on the extracted customer question items, the search queries as recorded in the click action log data indicates as useless, the useless as indication of the search queries is based on an interactive operation of answering whether at least a part of question items of FAQ in search results of the search queries has been determined as helpful for identifying responses to the customer question items, and the search queries are distinct from the question items of FAQ;
• performing recursively clustering of items for identifying similarity among the question items of FAQ and search queries for retrieving the question items of FAQ, the items comprising at least one of:
• the question items of FAQ registered in the FAQ search system,
• the customer question items, and
• the extracted search queries;
• and
• based on the clustered items, providing at least one candidate question item of FAQ for maintenance, to cause either add or delete an FAQ associated with the at least one candidate question item of FAQ in the FAQ search system.

A computer-readable non-transitory recording medium storing computer-executable instructions that when executed by a processor cause a computer system to:

• extract, from reception log data, customer question items, the customer questions representing question items posed by a customer, from contents of an inquiry from the customer and a response to the customer, wherein the extracting of the customer question items deletes customer question items that are not associated with the search queries among the extracted customer question items;
• extract, from click action log data, search queries previously executed to search for question items of FAQ in an FAQ search system, wherein the search queries are based on the extracted customer question items, the search queries as recorded in the click action log data indicates as useless, the useless as indication of the search queries is based on an interactive operation of receiving an answer whether at least a part of question items of FAQ in search results of the search queries has been determined as helpful for identifying responses to the customer question items, and the search queries are distinct from the question items of FAQ;
• perform recursively clustering of items for identifying similarity among the question items of FAQ and search queries for retrieving the question items of FAQ, the items comprising at least one of:
• the question items of FAQ registered in the FAQ search system,
• the customer question items, and
• the extracted search queries;
• and
• based on the clustered items, provide at least one candidate question item of FAQ for maintenance to cause either add or delete an FAQ associated with the at least one candidate question item of FAQ in the FAQ search system.