JP7456179B2 - Search system, search result presentation method, and search result presentation program - Google Patents

Description

The present invention relates to a technique for presenting information search results.

In order to enable customers to resolve problems and questions regarding products and services by themselves, we provide FAQs (Frequently Asked Questions) based on past inquiries.
Asked Questions) is being developed in many companies. When a customer actually encounters a problem, they can browse the FAQ and search for the information they need.

Several techniques are known to improve the convenience of this FAQ.

For example, when a user selects a business item, a help screen that includes a first screen area corresponding to a subordinate business item of the business item and a second screen area corresponding to a derived business item of the business item. A technique for displaying a screen is known (for example, see Patent Document 1).

Furthermore, for example, there is a known technology that extracts corresponding questions based on search conditions specified by an operator from a FAQ database in which related keywords are associated with each of multiple categories that are a cross-sectional classification of multiple questions. (For example, see Patent Document 2).

JP 2019-8663 Publication JP2019-28898A

Such FAQs may not be used effectively. For example, even though the FAQ contains the information necessary to solve a problem, a customer may inquire over the phone. One possible reason for this is the ``complexity'' of the FAQ search result list screen.

For example, a customer views the details of the FAQ by relying on the items displayed as the top page of the FAQ on the screen of a web browser and tracing lower-level items related to the customer. The customer can also enter a search query to search for necessary FAQs. However, in the case of such a FAQ search, all hit combinations of "Q" (question content) and "A" (answer content) are simply displayed as a list as search results. For this reason, it is difficult for customers to quickly find FAQs related to themselves from among a large amount of text information. As a result, users may abandon searching and viewing the FAQ midway through, even though the FAQ contains information useful for solving a problem. This can lead to a decrease in customer satisfaction and an increase in the number of telephone inquiries to the call center.

In one aspect, an object of the present invention is to provide a search result list display screen that makes it easy to find a desired search result.

According to one embodiment, a search system includes a reception section, a search section, a calculation section, a screen creation section, and an output section. The reception unit receives input of a search query including a search keyword. The search unit searches for a document in which any one or more of a plurality of item information having a hierarchical structure of at least three layers is associated according to the content of the document , and between each layer in the hierarchical structure. A document search based on a search query is performed on a storage unit that stores a plurality of documents that are associated with each piece of item information that has a series of parent-child relationships . The calculation unit calculates, for each search result document that is a document obtained as a result of the search, a degree of similarity representing the degree of similarity of the search result document to the search query. The screen creation unit is a screen that displays a list of search result documents, and displays the highest search result document with the highest degree of similarity among the search result documents, and each document in the hierarchical structure associated with the highest search result document. A first related search result document, which is a search result document to which all item information having a series of parent-child relationships between hierarchies is associated, is included in the first group, which is a search result document. of the remaining search result documents, and of the item information associated with the highest search result document among the remaining search result documents, information about the lowest level in the hierarchical structure. Create a screen that displays a second related search result document to which all of the item information except for is associated so that it can be further distinguished from other search result documents other than the second related search result document among the remaining search result documents. . The output unit outputs the created screen.

According to the above aspect, it is possible to provide a search result list display screen that makes it easy to find the desired search result.

1 is a diagram illustrating an example of a functional configuration of a search system. It is a diagram showing an example of the data structure of FAQDB. FIG. 2 illustrates an example of a data structure of an item information DB. FIG. 3 is a diagram showing a table (part 1) showing an example of the correspondence between the search results of the search unit, the calculation results of the calculation unit, and each item information. FIG. 3 is a diagram illustrating an example (part 1) of a search result display screen. It is a figure showing the table (part 2) showing the example of the correspondence of the search result of a search part, the calculation result of a calculation part, and each item information. FIG. 7 is a diagram illustrating an example (part 2) of a search result display screen. 1 is a diagram illustrating an example of a hardware configuration of an information processing device. 12 is a flowchart (Part 1) showing an example of FAQ service provision processing. 12 is a flowchart (Part 2) illustrating an example of FAQ service provision processing. 12 is a flowchart (part 3) illustrating an example of FAQ service provision processing. 12 is a flowchart (part 1) showing an example of search result FAQ arrangement processing. 12 is a flowchart (part 2) showing an example of search result FAQ arrangement processing.

Hereinafter, embodiments will be described in detail with reference to the drawings.

Figure 1 shows an example of the functional configuration of a search system.

The search system 10 is connected to a terminal device 1 operated by a user of the search system 10 via a communication network 2 such as the Internet, and data cannot be exchanged with the terminal device 1. can.

The search system 10 includes a reception section 11 , a search section 12 , a calculation section 13 , a screen creation section 14 , an output section 15 , and a storage section 16 .

The receiving unit 11 receives input of a search query sent from the terminal device 1 via the communication network 2 . The search query is created by the user and input into the terminal device 1 in order to obtain a desired FAQ. This search query includes a search keyword.

The search unit 12 searches the storage unit 16 for documents based on the search query received by the reception unit 11 .

The storage unit 16 stores a plurality of document data. This document is associated with any one or more of a plurality of item information depending on its contents.

In this embodiment, the storage unit 16 includes a FAQDB (database) 20 and an item information DB 30. The FAQDB 20 stores a plurality of data regarding FAQs. This FAQ is a document in which a sentence expressing the content of a question is associated with a sentence expressing the content of the answer to the question. Moreover, each item information is stored in the item information DB 30 in association with each FAQ.

The FAQ DB 20 and item information DB 30 will be further explained. FIG. 2 shows an example of the data structure of the FAQ DB 20, and FIG. 3 shows an example of the data structure of the item information DB 30.

The FAQ DB 20 has an FAQ table 20a and a keyword table 20b.

In the FAQ table 20a, "FAQID" is associated with "question" and "answer." In the FAQ table 20a, the "FAQID" column stores FAQID, which is identification information for identifying each FAQ. Further, the "Question" column and the "Answer" column store sentences expressing the content of the question and the content of the answer, respectively, for the FAQ specified by the FAQ ID.

In the keyword table 20b, "keyword" and "FAQID" are associated. FAQ keywords are stored in the "Keyword" column. FAQ keywords are words and phrases extracted from questions and answers regarding FAQs. The "FAQID" column stores at least one FAQ ID of the FAQ that includes the extracted FAQ keyword shown in the "Keyword" column.

In this embodiment, keywords are extracted from each FAQ text stored in the FAQ table 20a by using a processing method such as morphological analysis or N-gram to extract character strings to be used as FAQ keywords. It is assumed that the table 20b is created in advance.

The search unit 12 searches the FAQDB 20 for FAQs that include FAQ keywords that match the search keywords included in the search query. In this search, more specifically, the search unit 12 first searches the keyword table 20b for an FAQ keyword that matches the search keyword, and obtains the FAQ ID associated with the FAQ keyword. . Next, the search unit 12 refers to the FAQ table 20a and obtains the texts of the FAQ questions and answers associated with each of the obtained FAQ IDs. That is, the keyword table 20b provides a function as a transposed index for the FAQ sentences stored in the FAQ table 20a.

In the item information DB 30, "large items", "medium items", and "small items" are associated with "FAQID".

In this embodiment, each FAQ is categorized by its creator according to the content of the question and answer. Furthermore, each item in this categorization has a three-level hierarchical structure: upper, middle, and lower. Item information is identification information that identifies each of these items. In this embodiment, the item name, which is the name given to each item, is used as item information.

The "Major Item" column, "Medium Item" column, and "Small Item" column all store item names. The "Major Item" column stores item names for items at higher levels in the hierarchical structure described above. The "Medium Item" column stores item names for items at middle levels in the hierarchical structure described above. And the "Small Item" column stores item names for items at lower levels in the hierarchical structure described above.

Furthermore, the "FAQID" column stores at least one FAQ ID that specifies the FAQ stored in the FAQ table 20a in the FAQDB 20.

In the item information DB 30, item names of large items are associated with one or more item names of medium items, and item names of one or more small items are associated with item names of medium items. That is, in the item information DB 30, a tree structure formed by the mutual relationships between large items, medium items, and small items is defined.

In addition, in the item information DB 30, item names of large items, medium items, and small items are associated with each FAQ specified by the FAQ ID stored in the "FAQID" column. In other words, the item information DB 30 represents the association between each FAQ and each item name that has a series of parent-child relationships between each layer in the above-described hierarchical structure.

The explanation of FIG. 1 will be continued.

The calculation unit 13 calculates the degree of similarity for each of the search result documents. The search result document is a document obtained as a result of the search by the search unit 12, and the degree of similarity represents the degree of similarity of the search result document to the search query. In this embodiment, the calculation unit 13 calculates the degree of similarity for the search result document based on the search keyword included in the search query using BM25, which is a well-known method. Note that instead of BM25, the similarity may be calculated using other known methods such as TF-IDF or cosine similarity.

The screen creation unit 14 creates a screen that displays the results of the search by the search unit 12.

The output unit 15 outputs the screen created by the screen creation unit 14. The output screen is sent to the terminal device 1 via the communication network 2 and displayed on the terminal device 1.

Here, the creation of a screen for displaying the search results by the search unit 12, which is performed by the screen creation unit 14, will be further explained using FIGS. 4 to 7.

FIG. 4 is a table (part 1) showing an example of the correspondence between the search results of the search unit 12, the calculation results of the calculation unit 13, and each item information.

The portion of this table labeled A is provided with a "FAQ" column and a "similarity" column. The "FAQ" column shows identification information about the FAQ, which is a search result document obtained by the search by the search unit 12, and the "similarity" column shows the identification information about the FAQ, which is a search result document obtained by the search by the search unit 12. The calculated similarity value is shown. Further, in the part marked B in this table, columns of "large item", "medium item", and "small item" are provided. The "Major item", "Medium item", and "Small item" columns contain the large item, medium item, and Identification information that specifies item information for each item and sub-item is displayed.

In the table of FIG. 4, for example, pay attention to the top row in which "FAQ4-1-1-2" is displayed in the "FAQ" column. It can be seen that the FAQ specified by "FAQ4-1-1-2" had a similarity of "31.52". Furthermore, it can be seen that each piece of item information specified by "Major Item 4," "Medium Item 4-1," and "Small Item 4-1-1" was associated with this FAQ.

As can be seen from the part marked A in FIG. 4, in this table, the FAQs obtained as a result of the search by the search unit 12 are arranged in descending order of the similarity value calculated by the calculation unit 13, i.e. , arranged in descending order of similarity. This arrangement has nothing to do with the item information, and as can be seen from the part labeled B in FIG. 4, the FAQs are not classified by item.

FIG. 5 is an example (part 1) of a search result display screen.

The display screen 50 shown in FIG. 5 displays the "Q" (text expressing the content of the question) and "A" (text expressing the content of the answer) in each FAQ in the order of the FAQs in the table of FIG. A list is displayed in order. On this display screen 50, the FAQs are arranged in descending order of similarity to the search query, but the user cannot always create the most appropriate search query to search for the desired FAQ. . For this reason, it is not uncommon for a desired FAQ to be displayed buried in a lower position in such a list display, and in such cases, it is not easy for the user to find the desired FAQ. In many cases when users browse FAQs, they want a quick solution to their problem. Therefore, displaying a large amount of text information as a list as search results in this manner is not optimal.

Therefore, the screen creation unit 14 creates a screen that displays a list of FAQ search results as described below.

First, the screen creation unit 14 identifies the highest search result document from the search result documents obtained as a result of the search by the search unit 12. The highest search result document is the document with the highest degree of similarity calculated by the calculation unit 13 among the search result documents.

In the table illustrated in FIG. 4, among the FAQs that are the search results by the search unit 12, the similarity for "FAQ4-1-1-2" is "31.52", and the value of this similarity is "FAQ4-1-1-2". has the highest degree of similarity among the FAQs that are the search results. Therefore, this "FAQ4-1-1-2" is specified as the highest search result document.

Next, the screen creation unit 14 creates a screen that displays a list of search result documents. However, on this screen, the display of search result documents included in the first group, which is a group formed based on the highest search result document, can be distinguished from the display of the remaining search result documents. indicate. Here, the first group is a group that includes the highest search result document and the first related search result document, and the first related search result document is the one in which all the item information associated with the highest search result document is associated. This is the search result document.

In the table illustrated in Figure 4, "FAQ4-1-1-2", which is the highest search result document, contains "Major Item 4" and "Major Item 4", which have a series of parent-child relationships in the hierarchical structure of item information Item information of ``Medium Item 4-1'' and ``Small Item 4-1-1'' are associated. In this table, the FAQ to which all of these item information are associated is "FAQ4-1-1-1." Therefore, in the table illustrated in FIG. 4, since the first related search result document is "FAQ4-1-1-1", the FAQ included in the first group is "FAQ4-1-1-2". and “FAQ4-1-1-1”. On the list display screen created by the screen creation unit 14, “FAQ4-1-1-2” and “FAQ4-1-1-1” among the FAQs that are the search results by the search unit 12 are replaced with the remaining FAQs. be displayed in an identifiable manner.

The highest search result document has the highest degree of similarity to the search query created by the user, and is therefore the search result document that is estimated to be most likely to be desired by the user. In addition, since the first related search result document is a search result document to which all of the item information associated with the highest search result document is associated, it is assumed that the first related search result document is extremely highly related to the highest search result document. This is the result document. Therefore, it is estimated that the possibility that the search result document desired by the user is the first related search result document is quite high. However, the first related search result document may have a low degree of similarity to the search query created by the user, and in this case, the first related search result document may be buried in the search result document list display screen. There is. It is difficult for the user to immediately find the first related search result document from such a list display screen.

On the other hand, on the search result document list display screen created by the screen creation unit 14, the highest search result document and the first related search result document included in the first group are displayed so as to be distinguishable from the remaining search result documents. be done. Therefore, even if the desired search result document is the first related search result document, the user can easily find the desired search result document.

Note that the screen creation unit 14 creates a list display screen that displays the search result documents for the first group in order of similarity for each of the highest search result document and the first related search result document. You may also create one.

Creation of this list display screen will be explained using the data example of the table in FIG. 4.

As described above, in this table, the FAQs included in the first group are "FAQ4-1-1-2" and "FAQ4-1-1-1." Here, the similarity of "FAQ4-1-1-2" is "31.52". On the other hand, the similarity of "FAQ4-1-1-1" is "17.50", which is smaller than the similarity of "FAQ4-1-1-2".

In this case, the screen creation unit 14 creates a list display screen in which the FAQs are arranged in order of similarity, that is, "FAQ4-1-1-2" and "FAQ4-1-1-1." You may also do so. On the list display screen created in this way, the FAQs are arranged in order of the likelihood that the document is the search result document desired by the user, making it easier for the user to find the search result document desired by the user.

In addition, the screen creation unit 14 displays the second related search result document among the remaining search result documents described above so that it can be further distinguished from the second related search result document among the remaining search result documents. A list display screen of search result documents may be created as shown in FIG. Here, the second related search result document means that all of the item information associated with the highest search result document except for the lowest layer in the hierarchical structure of item information is associated with the second related search result document. This is the search result document.

Creation of this list display screen will be explained using the data example of the table in FIG. 4.

In this table, among the item information associated with "FAQ4-1-1-2", which is the highest search result document, the item information for the lowest hierarchy is "Small item 4-1-1", Other item information is "major item 4" and "medium item 4-1." Furthermore, the remaining FAQs among the FAQs that are the results of the search by the search unit 12 are other than "FAQ4-1-1-2" and "FAQ4-1-1-1." Among these remaining FAQs, the FAQs to which "Major Item 4" and "Medium Item 4-1" are both associated are "FAQ4-1-3-2", "FAQ4-1-2-2", "FAQ4-1-2-3" and "FAQ4-1-2-1", which are the second related search result documents.

In this case, the screen creation unit 14 selects the FAQs of these second related search result documents among the remaining FAQs other than "FAQ4-1-1-2" and "FAQ4-1-1-1". A list display screen may be created that displays the FAQ, which is not the second related search result document, in a more distinguishable manner.

As mentioned above, the second related search result document is one in which all of the item information other than the lowest level in the hierarchical structure of item information matches that of the highest search result document, so the second related search result document is the same as the first related search result document. It is estimated that it is highly related to the highest search result document after the search result document. Therefore, it is estimated that the possibility that the search result document desired by the user is the second related search result document is as high as that of the first related search result document. Therefore, by displaying the second related search result document among the remaining search result documents so that it can be distinguished from the second related search result document, the user can confirm that the desired search result document is the second related search result document. Even if it is a result document, it becomes easier to find the desired search result document.

In this case, the screen creation unit 14 displays second related search result documents in which the lowest level of associated item information is different from each other. You may also create a list display screen that displays the information in an identifiable manner.

Creation of this list display screen will also be explained using the data example of the table in FIG. 4.

In this table, the FAQs of the second related search result documents are "FAQ4-1-3-2", "FAQ4-1-2-2", "FAQ4-1-2-3", and This is "FAQ4-1-2-1". The item information of these lowest hierarchy is "Small item 4-1-3" for "FAQ4-1-3-2", while "FAQ4-1-2-2", "FAQ4-1- 2-3" and "FAQ 4-1-2-1" are "Small Item 4-1-2." In other words, the item information in the lowest hierarchy is different between the two.

In this case, the screen creation unit 14 creates the FAQ for "FAQ4-1-3-2", "FAQ4-1-2-2", "FAQ4-1-2-3", and "FAQ4-1- 2-1" may be created.

In this way, by displaying the second related search result documents in which the item information about the lowest hierarchy is different from each other in a distinguishable manner, item information about the lowest hierarchy from among the second related search result documents is displayed. The user can easily recognize the same items.

Further, in this case, the screen creation unit 14 classifies the second related search result documents into a plurality of second groups, and displays the second related search result documents included in each of the second groups between the second groups. A list display screen may be created in which each second group is arranged in a predetermined order so as to be distinguishable. Here, the second group is a group that includes second related search result documents that match the lowest level of associated item information. Further, the predetermined order is the order in which the maximum value of the degrees of similarity for each of the second related search result documents belonging to each of the second groups is large.

Creation of this list display screen will also be explained using the data example of the table in FIG. 4.

In this table, as mentioned above, the FAQs of the second related search result documents are "FAQ4-1-3-2", "FAQ4-1-2-2", "FAQ4-1-2-3", and This is "FAQ4-1-2-1". Among these, the item information at the lowest level for "FAQ4-1-3-2" is "Small item 4-1-3." In addition, the lowest level item information for "FAQ4-1-2-2", "FAQ4-1-2-3", and "FAQ4-1-2-1" is "Small item 4-1-2". ”. Therefore, in this table, the FAQ of the second related search result document is group A that includes only "FAQ4-1-3-2", "FAQ4-1-2-2", "FAQ4-1-2- 3" and Group B, which includes "FAQ4-1-2-1." These groups become the second group in the data example of FIG.

Here, the similarity of "FAQ4-1-3-2", "FAQ4-1-2-2", "FAQ4-1-2-3", and "FAQ4-1-2-1" is, respectively. They are "30.77", "29.01", "28.06", and "20.58". Therefore, the maximum value of similarity in group A is "30.77" and the maximum value of similarity in group B is "29.01". Therefore, the maximum value of similarity is greater for group A than for group B.

In this case, the screen creation unit 14 may create a list display screen that identifiably displays the FAQs included in group A and the FAQs included in group B in this order. .

In the list display screen created in this way, priority is given to displaying the second group containing the second related search result document that is more likely to be the search result document desired by the user. This makes it easier for users to find desired search result documents.

Further, in this case, the screen creation unit 14 displays the second related search result document that matches the lowest hierarchy of the associated item information. A list display screen may be created in which the items are arranged and displayed in the order of the degree of similarity.

Creation of this list display screen will also be explained using the data example of the table in FIG. 4.

As mentioned above, Group B, which includes "FAQ4-1-2-2", "FAQ4-1-2-3", and "FAQ4-1-2-1", is the second group and is the lowest group. This is a group of second related search result documents including FAQs that have matching item information about the hierarchy. Here, the similarity of "FAQ4-1-2-2", "FAQ4-1-2-3", and "FAQ4-1-2-1" is "29.01" and "28.06", respectively. ”, and “20.58”.

In this case, the screen creation unit 14 displays the FAQs included in this group B in descending order of similarity: "FAQ4-1-2-2," "FAQ4-1-2-3," and "FAQ4." -1-2-1" may be created.

On the list display screen created in this way, FAQs with matching item information for the lowest hierarchy are arranged in order of likelihood that they are the search result documents desired by the user. Therefore, it becomes easier for the user to find the desired search result document.

Here, FIGS. 6 and 7 will be explained. FIG. 6 is a table (part 2) showing an example of the correspondence between the search results of the search unit 12, the calculation results of the calculation unit 13, and each item information. Moreover, FIG. 7 is an example (part 2) of a display screen of search results.

The table in FIG. 6 is exemplified in the table in FIG. 4 according to the order of FAQs based on item information and similarity on the FAQ list display screen created by the screen creation unit 14 as described above. This is a rearranged version of each FAQ. The display screen 70 shown in FIG. 7 displays the "Q" (text expressing the content of the question) and "A" (text expressing the content of the answer) in each FAQ in the order of the FAQs in the table of FIG. They are arranged in order and are created by the screen creation section 14.

As mentioned above, the FAQs included in the first group are "FAQ4-1-1-2", which is the highest search result document, and "FAQ4-1-1-1", which is the first related search result document. This is the FAQ. On the display screen 70, a blank line is inserted between the display of the FAQ included in the first group and the display of the remaining FAQ, and this blank line allows the display of the FAQ included in the first group and the remaining FAQ. The FAQ is clearly displayed. Also, as you can see from the part marked A in Figure 6, the FAQs for "FAQ4-1-1-2" and "FAQ4-1-1-1" are arranged in descending order of similarity. They are arranged on the display screen 70 in the following order.

Furthermore, as mentioned above, the second related search result document FAQ is "FAQ4-1-3-2", "FAQ4-1-2-2", "FAQ4-1-2-3", and "FAQ4-1-2-3". FAQ 4-1-2-1”. On the display screen 70, a blank line is also inserted between the display of the FAQ “FAQ4-1-2-1” and the display of the FAQ “FAQ3-1-1-2”. On the display screen 70, the blank line allows the FAQ, which is the second related search result document, and the FAQ, which is other than the second related search result document among the remaining FAQs mentioned above, to be displayed in a distinguishable manner. .

In addition, as mentioned above, there is a group A that includes only "FAQ4-1-3-2" and a group A that includes only "FAQ4-1-3-2", "FAQ4-1-2-2", "FAQ4-1-2-3", and "FAQ4-1-2". Group B including "-1" is the second group.

As can be seen from the part marked B in FIG. 4-1-3”. Furthermore, as can be seen from the part marked A in FIG. 6, the degree of similarity of "FAQ4-1-3-2" with respect to FAQ is "30.77". On the other hand, the lowest hierarchy for each FAQ of "FAQ4-1-2-2", "FAQ4-1-2-3", and "FAQ4-1-2-1" among the second related search result documents The item information is "Small item 4-1-2". The similarity for each FAQ "FAQ4-1-2-2", "FAQ4-1-2-3", and "FAQ4-1-2-1" is "29.01" and "FAQ4-1-2-1," respectively. 28.06'' and ``20.58'', therefore, the maximum value of similarity is ``29.01''.

On the display screen 70, a blank line is also inserted between the display of the FAQ "FAQ4-1-3-2" and the display of the FAQ "FAQ4-1-2-2." On the display screen 70, the blank lines are used to distinguishably display FAQs for groups A and B, which differ in the lowest level of associated item information. Furthermore, on the display screen 70, the FAQ “FAQ4-1-3-2” is displayed before (at the top) the FAQ “FAQ4-1-2-2”. This display displays the second groups in descending order of the maximum similarity for each of the second related search result documents belonging to each second group. Furthermore, on the display screen 70, the FAQs "FAQ4-1-2-2", "FAQ4-1-2-3", and "FAQ4-1-2-1" included in group B are displayed in this order. are displayed in order. This display shows the display of the second related search result documents that match the lowest hierarchy of the associated item information, and the display of the high degree of similarity for the second related search result documents. They are displayed in order.

Next, FIG. 8 will be explained. FIG. 8 shows an example of the hardware configuration of an information processing device (computer).

This information processing device 80 can be used as the search system 10 in FIG.

The information processing device 80 includes a CPU (Central Processing Unit) 81, a memory 82, an input device 83, an output device 84, an auxiliary storage device 85, a medium drive device 86, and a network connection device 87. These components are hardware and are connected to each other by a bus 88.

The memory 82 is, for example, a semiconductor memory such as a ROM (Read Only Memory), a RAM (Random Access Memory), or a flash memory, and stores programs and data used for processing.

The CPU 81 (processor) provides the functions of the search system 10 by executing programs using the memory 82, for example. Therefore, the CPU 81 can operate as the reception section 11, search section 12, calculation section 13, screen creation section 14, and output section 15 in FIG. 1 by executing programs using the memory 82.

The input device 83 is, for example, a keyboard, a pointing device, or the like, and is used to input instructions or information from the administrator of the search system 10. The output device 84 is, for example, a display device, a printer, a speaker, etc., and is used for inquiries or instructions from the administrator and for outputting processing results.

The auxiliary storage device 85 is, for example, a magnetic disk device, an optical disk device, a magneto-optical disk device, a tape device, or the like. Auxiliary storage device 85 may be a hard disk drive or flash memory. The information processing device 80 can store programs and data in the auxiliary storage device 85 and load them into the memory 82 for use. Further, the auxiliary storage device 85 can also be used as the storage unit 16 in FIG. 1, and therefore can also be used as each of the FAQ DB 20 and the item information DB 30.

The medium drive device 86 drives the portable recording medium 89 and accesses the recorded contents. The portable recording medium 89 is a memory device, a flexible disk, an optical disk, a magneto-optical disk, or the like. The portable recording medium 89 may be a CD-ROM (Compact Disk Read Only Memory), a DVD (Digital Versatile Disk), a USB (Universal Serial Bus) memory, or the like. The administrator of the search system 10 can store programs and data in the portable recording medium 89 and load them into the memory 82 for use.

In this way, a computer-readable recording medium that stores programs and data used in processing is a physical (non-temporary) recording medium such as the memory 82, the auxiliary storage device 85, or the portable recording medium 89. It is a medium.

The network connection device 87 is a communication interface circuit that is connected to a communication network, such as a LAN (Local Area Network) or a WAN (Wide Area Network), including the communication network 2, and performs data conversion associated with communication. The information processing device 80 can also receive programs and data from an external device via the network connection device 87 and load them into the memory 82 for use.

Note that the information processing device 80 does not need to include all of the components shown in FIG. 8, and some components may be omitted depending on the application or conditions. For example, if the portable recording medium 89 is not used, the medium drive device 86 may be omitted.

Next, the processing that the program causes the CPU 81 of the information processing device 80 to perform when executed by the CPU 81 will be described with reference to Figures 9 to 13.

FIG. 9, FIG. 10, and FIG. 11 are flowcharts showing examples of FAQ service provision processing. This process is a process for causing the information processing device 80 to provide the functions of the search system 10 in FIG. This is the process when the

This FAQ service process is started when it is detected that the network connection device 87 has received a search query created by the user, which is sent from the terminal device 1 via the communication network 2 .

When the FAQ service providing process is started, first, in S101 of FIG. 9, the CPU 81 performs a process of acquiring this search query. When the CPU 81 performs this process, the information processing device 80 provides the function of the reception unit 11 in FIG.

Next, in S102, the CPU 81 performs a process of searching the FAQ DB 20 stored in the auxiliary storage device 85 for FAQs based on the acquired search query, and acquiring the search results. In this process, the FAQ ID of the FAQ in which the keyword included in the search query was included as an FAQ keyword in the "Keyword" column in the keyword table 20b of the FAQDB 20 shown in FIG. 2 is acquired. The FAQ obtained through the process of S102 will be referred to as a "search result FAQ." When the CPU 81 performs this process, the information processing device 80 provides the function of the search unit 12 in FIG.

Next, in S103, the CPU 81 performs a process of calculating the degree of similarity of each of the obtained search results FAQ to the search query obtained in the process of S101. In this process, sentences of FAQ questions and answers specified by the FAQ ID for each of the obtained search result FAQs are acquired from the FAQ table 20a of the FAQDB 20, and the degree of similarity of the obtained sentences to the search query is calculated. be done. When the CPU 81 performs the process of S103, the information processing device 80 provides the function of the search unit 12 in FIG.

Next, in S104, the CPU 81 performs a process of acquiring item information for each of the obtained search result FAQs from the item information DB 30 stored in the auxiliary storage device 85. In this process, item information (item information for major items, medium items, and minor items) associated with the FAQ ID for each of the obtained search result FAQs is acquired.

The subsequent processing from S105 to S119 in FIG. 11 is a process for creating a search result FAQ list display screen, such as the display screen 70 in FIG. When the CPU 81 performs these processes, the information processing device 80 provides the function of the screen creation section 14 in FIG. 1.

First, in S105, the CPU 81 performs a process of identifying the search result FAQ that has the maximum similarity value calculated in the process of S103 among those that have not yet been placed on the list display screen to be created. will do. The search result FAQ specified through this process will be referred to as a "target FAQ." The target FAQ specified by the process of S105 corresponds to the above-mentioned highest search result document.

Next, in S106, the CPU 81 performs search result FAQ arrangement processing. This process is a process of arranging search result FAQs whose item information at both the upper and middle levels match the target FAQ on the list display screen to be created.

If the target FAQ is the highest search result document, the search result FAQ arrangement process in S106 displays the search result FAQs, which are the first related search result document and the second related search result document, together with the target FAQ on a list display screen. The process is to place it in The details of this search result FAQ placement process will be described later.

Next, in S107, the CPU 81 performs a process of determining whether or not there are any search results FAQ obtained in the process of S102 that are not placed on the list display screen being created. In this determination process, when the CPU 81 determines that unplaced search result FAQs remain (when the determination result is YES), the CPU 81 advances the process to S108. On the other hand, when the CPU 81 determines in this determination process that there are no unplaced search result FAQs remaining (when the determination result is NO), the CPU 81 advances the process to S120 in FIG. 11.

Next, in S108, the CPU 81 performs a process of extracting, from among the search result FAQs that remain unarranged, those whose item information in the upper hierarchy matches the target FAQ.

As described above, in the search result FAQ placement process performed as the process of S106, search result FAQs whose item information in both the upper and middle hierarchies matches the target FAQ are placed on the list display screen. Therefore, the process of S108 extracts search result FAQs whose item information in the upper hierarchies matches the target FAQ (i.e., the search result FAQ that is the best search result document) but whose item information in the middle hierarchies does not match the target FAQ.

When the CPU 81 completes processing of S108, the process proceeds to S109 in FIG. 10.

Next, in S109, the CPU 81 performs a process of determining whether or not the search result FAQ extracted by the extraction process of S108 exists. In this determination process, when the CPU 81 determines that there is a search result FAQ to be extracted (when the determination result is YES), the CPU 81 advances the process to S110. On the other hand, in this determination process, when the CPU 81 determines that there is no search result FAQ to be extracted (when the determination result is NO), the CPU 81 advances the process to S115.

Next, in S110, among the search result FAQs extracted by the extraction process in S108, the one with the highest degree of similarity calculated in the process in S103 is identified, and the identified search result FAQ is newly set as the target FAQ. The CPU 81 performs the setting process.

Through the process of S110, the similarity is determined among the search results FAQ, which is the highest search result document, and the item information of the upper hierarchy matches, but the item information of the middle hierarchy does not match. The one with the largest is set as the target FAQ.

Next, in S111, the CPU 81 performs a process of arranging a blank line by inserting a line break on the list display screen being created. By placing this blank line, the search result FAQ placed on the list display screen by the process executed before this process and the search result FAQ placed on the list display screen by the process executed after this process. identification becomes possible. Next, in S112, the CPU 81 performs the same search result FAQ arrangement process as that performed in S106. However, in the process of S112, the target FAQ is different from that in the process of S106.

Next, in S113, the CPU 81 performs a process of determining whether or not there are any search result FAQs extracted by the extraction process of S108 that have not been placed on the list display screen being created. In this determination process, when the CPU 81 determines that unplaced search result FAQs remain (when the determination result is YES), the CPU 81 advances the process to S114. On the other hand, in this determination process, when the CPU 81 determines that there are no unplaced search result FAQs remaining (when the determination result is NO), the CPU 81 advances the process to S115.

Next, in S114, among the extracted search result FAQs that remain unplaced on the list display screen, the one with the highest degree of similarity is identified, and the identified search result FAQ is , the CPU 81 performs a process of setting it as a target FAQ again. After completing the process of S114, the CPU 81 returns the process to S111, and re-executes the process from S111 onwards with respect to the newly set target FAQ.

By repeating the above processing from S110 to S114, the search FAQ, which is the highest search result document, matches the item information in the upper hierarchy, but the item information in the middle hierarchy does not match. is placed on the list display screen.

On the other hand, if the result of the determination process in S109 or S113 is NO, in S115, the CPU 81 performs a process of determining whether or not any of the search result FAQs acquired in the process of S102 remain unplaced on the list display screen being created. When the CPU 81 determines in this determination process that there are unplaced search result FAQs (when the determination result is YES), the process proceeds to S116 in FIG. 11. On the other hand, when the CPU 81 determines in this determination process that there are no unplaced search result FAQs (when the determination result is NO), the process proceeds to S120 in FIG. 11.

The process from S116 to S119 in FIG. 11 is a process for arranging the remaining search result FAQs, which are extracted by the extraction process in S108 and whose item information in a higher hierarchy does not match the search result FAQ that is the best search result document, on the list display screen.

First, in S116, among the search result FAQs that have not yet been placed on the list display screen, the one with the maximum similarity value calculated in the process of S103 is specified, and the specified search result FAQ is displayed again. The CPU 81 performs the process of setting as the target FAQ.

Next, in S117, the CPU 81 performs a process of arranging a blank line by performing a line break on the list display screen being created. By placing this blank line, the search result FAQ placed on the list display screen by the process executed before this process and the search result FAQ placed on the list display screen by the process executed after this process. identification becomes possible.

Next, in S118, the CPU 81 performs the same search result FAQ arrangement process as that performed in S106 and S112. However, even in this process of S118, the target FAQ is different from those in the processes of S106 and S112.

Next, in S119, the CPU 81 performs a process of determining whether or not there are still search result FAQs that have not been placed on the list display screen being created. In this determination process, when the CPU 81 determines that there are unplaced search result FAQs remaining (when the determination result is YES), the CPU 81 returns the process to S116, and returns the process to S116 for the unplaced search result FAQs. The processing from to S118 is performed. On the other hand, in this determination process, when the CPU 81 determines that there are no unplaced search result FAQs remaining (when the determination result is NO), the CPU 81 advances the process to S120.

Next, in S120, the CPU 81 outputs the search result FAQ list display screen created by the above processing, and controls the network connection device 87 to send it to the terminal device 1. When the CPU 81 performs this processing, the information processing device 80 provides the function of the output unit 15 in FIG. Thereafter, the CPU 81 ends this FAQ service providing process.

The process described above is the FAQ service providing process.

Next, the flowcharts shown in FIGS. 12 and 13 will be explained. This flowchart shows the processing contents of the search result FAQ placement process, which is executed by the CPU 81 as the processes of S106 (FIG. 9), S112 (FIG. 10), and S118 (FIG. 11) in the FAQ service provision process described above. It is. As mentioned above, this search result FAQ placement process places search result FAQs whose item information in both the upper and middle levels match the target FAQ on the list display screen to be created. It is processing.

When this search result FAQ arrangement process is started, first, in S201, the CPU 81 arranges the target FAQ on the list display screen, and then performs a line break process. In this process, the questions and answers associated with the FAQ ID for the search result FAQ, which is the target FAQ, are read from the FAQ table 20a of the FAQDB 20, are associated with each other on the list display screen, and are arranged in one line. Line feed processing is performed.

Next, in S202, among the search result FAQs that are not placed on the list display screen when this process is executed, the item information of the upper, middle, and lower layers are all the target FAQs. The CPU 81 performs processing to extract items that match the FAQ. Note that the item information for each search result FAQ is acquired from the item information DB 30 through the process of S104 (FIG. 9).

Next, in S203, the CPU 81 performs a process of determining whether the search result FAQ extracted by the extraction process of S202 exists. In this determination process, when the CPU 81 determines that there is a search result FAQ to be extracted (when the determination result is YES), the CPU 81 advances the process to S204. On the other hand, when the CPU 81 determines in this determination process that there is no search result FAQ to be extracted (when the determination result is NO), the CPU 81 skips the process of S204 and advances the process to S205.

Next, in S204, the CPU 81 arranges the search result FAQs extracted by the extraction process in S202 on the list display screen in order of similarity, and performs a process of starting a new line every time one search result FAQ is arranged. In the process of S204, the questions and answers associated with the FAQ ID for each of the extracted search result FAQs are read from the FAQ table 20a of the FAQDB 20, and are correlated on the list display screen and arranged line by line in order of similarity. A line break process is performed each time one line is placed. Note that the degree of similarity for each search result FAQ is calculated by the process of S103 (FIG. 9).

Through the processes of S201 and S204 described above, the search result FAQ, which is the target FAQ, and the search result FAQ in the order of similarity in which the item information of the upper, middle, and lower layers match the target FAQ are combined into one. They are collected and arranged on the list display screen. Therefore, when these processes are executed as the process of S106 (FIG. 9), the search results FAQ included in the first group, which is the group that includes the highest search result document and the first related search result document, as described above. are placed on the list display screen.

Next, in S205, the CPU 81 performs a process of determining whether or not there are any search results FAQ obtained in the process of S102 (FIG. 9) that are not placed on the list display screen. In this determination process, when the CPU 81 determines that unplaced search result FAQs remain (when the determination result is YES), the CPU 81 advances the process to S206. On the other hand, when the CPU 81 determines in this determination process that there are no unplaced search result FAQs remaining (when the determination result is NO), the CPU 81 ends this search result FAQ placement process and returns to the original process. .

Next, in S206, among the search result FAQs that are not placed on the list display screen at the time of execution of this process, the item information of each upper and middle hierarchy matches the search result FAQ that is the target FAQ. The CPU 81 performs the process of extracting what is to be done.

Next, in S207, the CPU 81 performs a process of determining whether or not the search result FAQ extracted by the extraction process of S206 exists. In this determination process, when the CPU 81 determines that there is a search result FAQ to be extracted (when the determination result is YES), the CPU 81 advances the process to S208 in FIG. On the other hand, when the CPU 81 determines in this determination process that there is no search result FAQ to be extracted (when the determination result is NO), the CPU 81 ends this search result FAQ placement process and returns to the original process. .

The processing from S208 to S215 in FIG. 13, which will be explained from now on, is to group the search result FAQs that match the search result FAQ whose item information in the upper and middle layers is the target FAQ into groups for each item information in the lower layer. This is a process of classifying and arranging each group on the list display screen. In this arrangement, each group is arranged in descending order of the maximum similarity value for the search result FAQs included in each group, and in order of descending similarity for the search result FAQs included in each group. . Therefore, when these processes are executed as the process of S106 (FIG. 9) in the FAQ service provision process, the above-mentioned search result FAQ included in the plurality of second groups in which the second related search result documents are classified are placed on the list display screen.

First, in S208, the CPU 81 performs a process of classifying the search result FAQs extracted by the extraction process in S206 into groups for each item information in the lower hierarchy.

Next, in S209, the CPU 81 performs a process of acquiring the number of groups created by the classification process in S208 and substituting it into variable G.

Next, in S210, the CPU 81 performs a process of acquiring, for each group, the maximum value of similarity for the search results FAQ included in the group created by the process of S208.

Next, in S211, the CPU 81 performs a process of assigning an initial value "1" to the variable k.

Next, in S212, the CPU 81 performs a process of arranging a blank line by performing a line break on the list display screen being created. By placing this blank line, the group of search results FAQ placed on the list display screen by the process executed before this process and the search results placed on the list display screen by the process executed after this process. It becomes possible to identify the FAQ group.

Next, in S213, the group with the k-th maximum similarity value is identified, the search result FAQs included in that group are arranged in the order of similarity on the list display screen, and one search result FAQ is displayed. The CPU 81 performs a process of starting a new line every time it is placed. In the process of S210, the questions and answers associated with the FAQ ID for each of the identified search result FAQs are read out from the FAQ table 20a of the FAQDB 20, and are correlated on the list display screen and arranged line by line in order of similarity. A line break process is performed each time one line is placed.

Next, in S214, the CPU 81 performs a process of adding "1" to the current value of variable k and assigning the resultant value to variable k anew.

Next, in S215, the CPU 81 performs a process of determining whether the value of the variable k after being changed by the process of S214 exceeds the value of the variable G, that is, the number of groups created by the process of S208. In this determination process, when the CPU 81 determines that the value of the variable k exceeds the value of the variable G (when the determination result is YES), the CPU 81 ends this search result FAQ placement process and returns to the original process. On the other hand, when the CPU 81 determines in this determination process that the value of the variable k does not exceed the value of the variable G (when the determination result is NO), the CPU 81 returns the process to S212 and displays the list display screen. Processes the placement of search result FAQs for groups that have not been placed.

The process described above is the search result FAQ placement process.

The information processing device 80 functions as the search system 10 by the CPU 81 executing the processes shown in each of the figures of FIGS. 9 to 13 above, and as a result of the FAQ search based on the search query created by the user, It becomes possible to provide a display screen 70 such as the one illustrated in FIG.

Although the disclosed embodiments and their advantages have been described in detail, it will be understood that those skilled in the art may make various modifications, additions, and omissions without departing from the scope of the present invention as expressly set forth in the claims.

For example, on the display screen 70 in FIG. 7, questions and answers are arranged in correspondence with each other as search results FAQ. Along with these items, item information for each hierarchy of large items, medium items, and small items, which is item information for the search result FAQ, is displayed in the form shown in the part marked with B in FIG. 6, for example. It may also be arranged and displayed in association with the question and answer. Further, a configuration may be adopted in which the user can switch between displaying and non-displaying all or part of these item information on the display screen 70.

In addition, in the disclosed embodiment, for each FAQ, the FAQ is based on the item information and the degree of similarity, targeting item information that has a hierarchical structure of three levels: upper, middle, and lower. A list display screen of search results FAQ is created according to the order of search results. Here, a screen displaying a list of search result FAQs according to the order of FAQs based on the item information and similarity, with item information that has a multi-level hierarchical structure associated with each FAQ. It is also possible to create

In addition, in the disclosed embodiment, a blank line is arranged between displaying FAQs on the search result FAQ list display screen as an identifiable display of FAQs. Alternatively, each FAQ may be displayed in a distinguishable manner by changing one or more of the display color, size, font, etc. of the FAQ.

1 Terminal device 2 Communication network 10 Search system 11 Reception section 12 Search section 13 Calculation section 14 Screen creation section 15 Output section 16 Storage section 20 FAQDB
20a FAQ table 20b Keyword table 30 Item information DB
50, 70 Display screen 80 Information processing device 81 CPU
82 Memory 83 Input device 84 Output device 85 Auxiliary storage device 86 Media drive device 87 Network connection device 88 Bus 89 Portable recording medium

Claims (7)

a reception unit that accepts input of search queries including search keywords;
The document in which any one or more of a plurality of item information having a hierarchical structure of at least three layers is associated according to the content of the document, and a series of parent and child items between each layer in the hierarchical structure. a search unit that searches for the documents based on the search query in a storage unit that stores a plurality of the documents to which related item information is associated ;
a calculation unit that calculates, for each search result document that is a document obtained as a result of the search, a degree of similarity representing the degree of similarity of the search result document to the search query;
A screen displaying a list of the search result documents, the screen displaying the highest search result document with the highest degree of similarity among the search result documents, and each of the documents in the hierarchical structure associated with the highest search result document. The items included in the first group that includes the first related search result document that is the search result document with which all of the item information that has a series of parent-child relationships between hierarchies are The highest search result document in the hierarchical structure of the item information associated with the highest search result document among the remaining search result documents is displayed so as to be distinguishable from the remaining search result documents among the search result documents. The second related search result document to which all of the item information except for the information regarding the lower hierarchy is associated can be further distinguished from the remaining search result documents other than the second related search result document. a screen creation unit that creates the screen to be displayed ;
an output unit that outputs the created screen;
A search system comprising: The screen creation unit is configured to arrange and display the search result documents for the first group in accordance with the order of the degree of similarity for each of the highest search result document and the first related search result document. The search system according to claim 1, wherein a screen is created. The screen creation unit displays the second related search result documents in which the lowest hierarchy among the associated item information is different from each other. The search system according to claim 1 or 2, wherein the screen is created to be displayed in an identifiable manner. The screen creation unit displays each of the second related search result documents in which the lowest hierarchy of the associated item information matches, as a display for the second related search result documents. are classified into a plurality of second groups including a plurality of second groups, the display of the second related search result documents included in each of the second groups is arranged for each second group so as to be distinguishable among the second groups, and Claim characterized in that the screen is created in which the two groups are arranged in descending order of the maximum value of the degrees of similarity for each of the second related search result documents belonging to each of the second groups. The search system described in 3 . The screen creation unit displays a display for a second related search result document in which the lowest hierarchy of the associated item information matches the second related search result document. 5. The search system according to claim 3 , wherein the screens are created to be displayed in order of similarity. Accepts input of search queries that include search keywords,
The document in which any one or more of a plurality of item information having a hierarchical structure of at least three layers is associated according to the content of the document, and a series of parent and child items between each layer in the hierarchical structure. performing a search for the documents based on the search query in a storage unit that stores a plurality of the documents to which related item information is associated ;
For each search result document that is a document obtained as a result of the search, calculate a degree of similarity representing the degree of similarity of the search result document to the search query,
A screen displaying a list of the search result documents, the screen displaying the highest search result document with the highest degree of similarity among the search result documents, and each of the documents in the hierarchical structure associated with the highest search result document. The items included in the first group that includes the first related search result document that is the search result document with which all of the item information that has a series of parent-child relationships between hierarchies are The highest search result document in the hierarchical structure of the item information associated with the highest search result document among the remaining search result documents is displayed so as to be distinguishable from the remaining search result documents among the search result documents. The second related search result document to which all of the item information except for the information regarding the lower hierarchy is associated can be further distinguished from the remaining search result documents other than the second related search result document. Create the above screen to display ,
Output the created screen,
A search result presentation method characterized in that processing is performed by a computer. Accepts input of search queries that include search keywords,
The document in which any one or more of a plurality of item information having a hierarchical structure of at least three layers is associated according to the content of the document, and a series of parent and child items between each layer in the hierarchical structure. performing a search for the documents based on the search query in a storage unit that stores a plurality of the documents to which related item information is associated ;
For each search result document that is a document obtained as a result of the search, calculate a degree of similarity representing the degree of similarity of the search result document to the search query,
A screen that displays a list of the search result documents, the highest search result document having the highest degree of similarity among the search result documents, and each of the documents in the hierarchical structure associated with the highest search result document. The items included in the first group that includes the first related search result document that is the search result document with which all of the item information that has a series of parent-child relationships between hierarchies are The highest search result document in the hierarchical structure of the item information associated with the highest search result document among the remaining search result documents is displayed so as to be distinguishable from the remaining search result documents among the search result documents. A second related search result document to which all of the item information except for information about lower layers is associated can be further distinguished from other search result documents other than the second related search result document among the remaining search result documents. Create the above screen to display ,
Output the created screen,
A search result presentation program that causes a computer to perform processing.

Images