JP2002117061A - Device and method for providing information - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an interactive information providing system based upon natural sentences. SOLUTION: A user is prompted to input a character string in natural sentence form, a prescribed keyword is obtained by performing prescribed natural language processing as to the character string inputted by the user, and a retrieval condition frame having prescribed retrieval items is generated according to the prescribed keyword, and retrieval from a 1st database is performed according to the generated retrieval condition frame and the retrieval result is presented to the user.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to an information providing apparatus and an information providing method, and more particularly, to a system for receiving a search condition from a user using WWW and providing a search result to the user.

[0002]

2. Description of the Related Art The Internet is composed of a network system and computers interconnected via the network system. Computers connected to each other exchange information by using various systems such as a mail system and a WWW system.

[0003] The WWW system is a system expressed as a client / server model, in which a graphical information resource (ie, a web server program on a client computer and a Web server program on a server site) is used. , Web pages) can be exchanged. WW
Information resources in the W system are URL (Uniform Resource
urce Locator). The user browses the Web page of the Web site on the Internet by using the URL of the information resource via a browser.
Make an acquisition request specifying. This acquisition request is transferred to the Web site specified by the URL. In response to the acquisition request, the Web site transmits the information resource to the user's client computer. Information resources transmitted from the Web site are displayed via a browser on the client computer. The WWW system also provides a hyperlink function. As a result, the user can access various information resources on the Internet with a simple operation. In addition, the WWW system uses CG
With the I (Common Gateway Interface) function, based on the request transmitted from the browser, it is possible not only to exchange information resources but also to perform more advanced processing.

For example, a search engine searches a database based on search conditions received from a user, and presents the search results to the user. If the user wishes to further narrow down the search results, he or she inputs search conditions. This allows the user to interactively search for information using the search engine.

[0005]

In a conventional search engine (information presentation system), a user must input complicated search conditions into an input form in order to obtain a desired search result. Was. In addition, since the input form is set on the Web site side, it cannot be said that all users have good usability. For example,
A user who has a lot of experience using the search system knows the habit of the search engine to a certain extent, so it is possible to input complicated search conditions according to the habit, but this results in a heavy input burden. Was supposed to be On the other hand, it is difficult for a user who has little experience in using a search engine to determine what search conditions to input, and the performance of the search engine cannot be sufficiently brought out.

Accordingly, an object of the present invention is to provide an interactive information presentation system based on natural sentences.

[0007]

According to a first aspect of the present invention, a keyword necessary for a search is obtained by performing a natural sentence dialog with a user, and a database is stored based on the obtained keyword. Searching and presenting the search results to the user.

In this case, according to the present invention, a predetermined scenario is prepared, and the dialogue with the user proceeds in accordance with the scenario. Further, the present invention increases the matching ratio with the database by performing a predetermined natural language process on a character string in a natural sentence format given in a dialogue with a user. That is, the present invention performs a morphological analysis and a syntactic analysis on a given character string to cut out a keyword, and when a cut-out keyword includes a fluctuation expression, the fluctuation is expressed so as to absorb the fluctuation expression. Convert keywords.

Further, according to the present invention, keywords embodied from the obtained keywords are obtained, and a search is performed in consideration of the relevance between these keywords. This means that even if the character string given by the user is a knowledge expression that is ambiguous in searching the database, it is gradually converted into a knowledge expression of a specific level. For this reason, the present invention establishes a category of viewpoints, and constructs a specialized knowledge database in which keywords of various attributes are set for each viewpoint. For example, in the case of a travel expertise database, "area"
Five viewpoints, “theme”, “hotel”, “famous place / historic site”, and “airline” are set. This allows
Even if the user gives the ambiguous expression "I want to visit archeological sites in Thailand", provide specific information such as "I will stay in Bangkok and go to Ayutthaya archeological site in Ayutthaya, a neighboring city" Will be able to do it.

Still further, according to the present invention, if there is no search result to be presented to the user as a result of searching the database according to the search condition (character string in a natural sentence format) given by the user, the user can re-enter the search condition. Without resetting the search conditions, search is performed again. As a result, it is possible to avoid presenting the user with 0 search results.

More specifically, the present invention urges a user to input a character string in a natural sentence format, performs predetermined natural language processing on the character string input by the user, and obtains a predetermined keyword. Creating a search condition frame (search condition data) having a predetermined search item based on the obtained predetermined keyword, searching the first database based on the created search condition frame,
An information providing method characterized by presenting the search result to the user. The search result is typically presented to a user performing an interactive operation, but is not limited to this. For example, the search result may be presented to another user registered in advance, or a predetermined The information content may be notified to the server and registered.

According to a second aspect of the present invention, a keyword necessary for a search is obtained by performing a natural sentence dialogue with a user, and a database is searched based on the obtained keyword. While presenting the search result to the user, if the user wishes to transmit a request or the like to the information source indicated in the search result, provide the information regarding the user as customer sending information to the information source. It is characterized by. The customer sending information includes the user's contact information (e.g., e-mail address), a character string input during the search, and a search condition character string determined based on the character string.

The present invention is also realized as a recording medium on which a program for causing a computer system to realize a predetermined function is recorded. Here, the computer system is preferably realized by a client / server model.

Further, the recording medium is, for example, a hard disk (HD), a DVD-RAM, a flexible disk (FD), a CD-ROM, etc.
Includes memory such as OM. The computer also includes, for example, a so-called microcomputer or the like in which a so-called central processing unit such as a CPU or an MPU performs a predetermined process by interpreting a program. The program is
The function may be realized by including a function of calling a function realized by a basic program such as an operating system.

[0015] In this specification, the term "means" does not simply mean a physical means, but also includes a case where the function of the means is realized by software. Further, the function of one unit may be realized by two or more physical units, or the function of two or more units may be realized by one physical unit.

[0016]

Next, embodiments of the present invention will be described with reference to the drawings.

[First Embodiment] FIG. 1 is a block diagram showing a schematic configuration of an information providing system according to the present embodiment. As shown in FIG. 1, the information providing system is typically realized as a client / server system using a WWW system on the Internet.

The server computer 1 includes a Web server program 11, a Web page 12, a CGI program (script) 13, and a database 14.
The Web server program 11 accepts an HTTP message sent from the user's client computer 2 and performs a process corresponding to the message.
For example, if the Web server program 11 determines that the message is for accessing the Web page specified by the URL, the Web server program 11 provides the Web page to the client computer 2 of the user. W
The eb server program 11 typically includes httpd
It is implemented in the server computer 1 as a daemon program called "."

The Web page 12 is document data that is designed using, for example, HTML or the like and that constitutes a screen to be provided to the user. In the present embodiment, the Web
As a page 12, W constituting a top screen and an interactive screen
An eb page is prepared. The top screen is for introducing a user first. The interactive screen is for prompting the user for an input for a search, accepting an input of information, and presenting a result according to the input information. This interactive screen can receive at least information in a natural sentence (free sentence) format. The Web page may be prepared in advance as static data, or may be dynamically generated by the CGI program 13 or the like. The CGI program 13 is such that the Web server program 11
This is a program executed when a specific message is received by P. The CGI program 13 is typically described in a programming language such as Perl or C ++. In the present embodiment, a program having a function as a search engine capable of providing information interactively with the user is prepared. Database 14 is a CGI
The program 13 stores various dictionaries and rules necessary for realizing interactive information provision.

FIG. 2 is a block diagram showing a functional configuration of the information providing system according to the present embodiment. Here, an information providing system for searching and providing tour information will be described as an example.

As shown in FIG. 1, the functional configuration on the server side in the information providing system includes a search engine unit 21 and a database unit 22. The search engine unit 21 includes a natural sentence analyzing unit 211 that analyzes information (character strings) in a natural sentence format sent from a user, a dialog control unit 212 that controls a dialog with the user, and a scenario prepared in advance. A scenario control unit 213 that executes an action along the search condition, a search condition generation unit 214 that generates a search condition according to the action executed by the scenario control unit 213, and a search condition generation unit 214 according to the action executed by the scenario control unit 213 A search condition execution unit 215 for executing the generated search condition is provided. Search condition execution unit 215
Is typically executed at the stage when all the search conditions are generated by the search condition generation unit 214.

FIG. 3 shows a natural sentence analyzer 2 according to the present embodiment.
11 is a block diagram for explaining the operation of FIG. In the figure, a morphological analysis unit 2111 performs a morphological analysis on an input natural sentence with reference to a morphological dictionary 22a. For example, when the natural sentence illustrated in FIG. 4A is input, the morphological analysis unit 2111 outputs the morphological analysis result illustrated in FIG. Syntax analyzer 2112
Performs a syntax analysis with reference to a syntax analysis table. For example, when the morphological analysis result shown in FIG. 4B is input, the syntax analyzer 2112 outputs the syntax analysis result shown in FIG. 4C. The qualified word correction processing unit 2113 refers to the qualified word correction rule table and applies the correction rule. Specifically, the qualified word correction processing unit 2113 focuses on the word (verb) associated with the qualified word, and corrects the fluctuation expression of the qualified word according to the correction rule. For example, FIG.
When the syntax analysis result shown in (c) is input, the modified word correction processing unit 34 outputs the correction result shown in FIG. In this example, the word "place" is corrected to the word "hotel" by paying attention to "stay (stop form" stay ")". The qualified word is used in the following qualified word extraction unit 2114. Modifier extractor 21
14 extracts a modifier from the correction result. For example,
When the correction result shown in FIG. 4D is input, the modifier extraction unit 2114 outputs each of the elements shown in FIG. 5A. Note that a tag (“start word position−end word position”) indicating the word position of the morphological analysis result character string is added to the element. The negative expression processing unit 2115 determines whether or not the extraction result obtained by the modifier extraction unit 2114 includes a negative expression. Convert to For example,
The negative expression processing unit 2115 converts the negative expression “charge is not expensive” shown in FIG. 5A into an affirmative expression “charge is expensive” as shown in FIG. 5 (b). Then, a negative tag “NOT” is added. The dictionary reference correction processing unit 2116 refers to the logical expression DB 22b and the morphological analysis result, converts the inflected form of the word to the basic form,
Add the part of speech. For example, “beautiful” shown in FIG. 5B is converted to “case particle is beautiful” as shown in FIG. 5C. The sentence pattern determination unit 2117 determines the sentence pattern of the entire analytic sentence composed of each element. In this example, a positive tag is added to the entire analysis sentence (FIG. 6).

FIG. 7 shows a dialogue control unit 21 according to this embodiment.
6 is a flowchart for explaining the operation of FIG. The dialogue control unit 212 converts the analysis result given from the natural sentence analysis unit 211 into an internal expression for search and converts the search result into an expression for user presentation. As shown in the figure, first, the dialogue control unit 212 performs an end determination process on the analysis result provided from the natural sentence analysis unit 211 with reference to the emotion dictionary DB 22c (STEP 701). FIG.
Is a diagram showing an example of the emotion dictionary DB 22c. When determining that there is a word indicating the intention to end, such as “stop”, the dialog control unit 212 ends the process (Yes in STEP 702). Dialogue control unit 212
Determines that there is no word indicating the intention to end (ST
Next, a question sentence determination process is performed (S702: No).
TEP703). When determining that the sentence is a question sentence (Yes in STEP 704), the dialog control unit 212 sets the execution example to the return value (STEP 705), and ends the process. The execution of the example means that the system executes a process of presenting an input example according to a scene in order to prompt the user for an input. For example, System: Where do you want to go? User: What to enter? In response to the dialogue, enter the system name: region, country, city, such as Asia, Thailand, and Bangkok. Set example execution. On the other hand, the dialog control unit 21
2 is when it is determined that the sentence is not a question (STEP 70
3), the word spotting process is performed with reference to the specialized field dictionary DB 22d (STEP 706). The word spotting process refers to extracting a character string (word) having a high degree of importance from an input character string of a natural sentence. In this example, character strings of elements (for example, destination, main purpose, departure date, number of people, etc.) regarding tour information are extracted. More specifically, the tour information element is an item necessary for searching for tour information pre-planned as a package tour, and is a fragmentary element of a travel plan requested by each user. If the dialog control unit 212 determines that spotting has failed (STEP 707)
No), set an example of presenting an option as a return value (ST
(EP708), the process ends. On the other hand, the dialog control unit 21
If it is determined that the spotting has succeeded (YES in STEP 707), the logical expression conversion process is performed with reference to the logical expression DB 22b (STEP 709).
FIG. 9 is a diagram illustrating an example of the logical expression DB 22b. The logical expression DB 22b is a rule base for converting descriptions such as “other than A” and “excluding A” into logical expressions such as “! A” suitable for internal processing. Then, the dialogue control unit 212 performs an expression conversion process with reference to the expression conversion DB 22e (STEP 710). FIG. 10 is a diagram illustrating an example of the expression conversion DB 22e. Expression conversion DB2
2e is a rule base for converting a variable expression such as “A / B” or “A month B day” into a unique expression.

FIG. 11 is a diagram for explaining a conceptual configuration of the specialized field dictionary DB 22d. Specialized field dictionary DB2
2d has a structure in which a plurality of data of “viewpoints” that absorb the polysemy with respect to the same “keyword” are linked in order to enable the system to handle homophonic polysemy. In this example, viewpoints of “area”, “purpose (theme)”, “hotel”, “famous place / historic site”, and “airline” are set. In addition, “abstract words”, “synonyms”, “related words”, “specific words”, and “explanatory sentences” are defined for each data belonging to each of these viewpoints. Among them, “abstract words”, “synonyms”, “related words” and “specific words” indicate zero or more keywords by pointers. To indicate zero keywords:
There is no keyword to refer to, and N
Pointing the keywords means that there are N keywords to be referred to. Furthermore, each keyword and between a keyword and a viewpoint are linked by “relevance” indicating the strength (weight) of semantic connection. FIG. 12 is a diagram for explaining a specific data structure of the specialized field dictionary DB 22d. As shown in the figure,
The specialized field dictionary DB 22d has a frame structure including slots and slot values. That is, FIG.
Indicates a link between keywords, and FIG.
Indicates this for each viewpoint, and indicates whether the keywords are an abstract relationship, a concrete relationship, a synonymous relationship, or a related relationship (a relationship across viewpoints). Among the relationships between keywords passing through viewpoints in FIG. 12, regarding the relationships of abstraction, concreteness, and synonymity, at least one of the viewpoints linked to the keyword of the link destination has the same type as the type of the viewpoint of the link source. It is required that there be. Specifically, when the type of the viewpoint of the link source is “area”, it is required that at least one of the viewpoints linked to the keyword of the link destination is “area”.

FIG. 13 is a flowchart for explaining the operation of the scenario control unit 213 according to the present embodiment. The scenario control unit 213 uses the user profile DB2 based on the processing result given from the dialog control unit 212.
With reference to 2f, a process for acquiring a search condition frame is performed (STEP 1301). That is, the scenario control unit 213 executes a scenario for acquiring an item that is not yet filled in the subsequent processing, for example, a question sentence “How many people will you go?” Or a search to execute a search. Get search condition frame from. Next, the scenario control unit 213 refers to the action rule base DB 22g and the scenario DB 22h,
An action extraction process is performed (STEP 1302). For example, in the case of a tour search, the action rule base DB 22g stores an area (destination) as shown in FIG.
Action rules on theme (purpose), schedule (departure date, return date, number of days), number of people, budget, etc. are if-the
It is defined in n format. In other words, it is defined that when the area is not determined, the action related to the area is fired (executed), and when the purpose is not determined, the action related to the purpose is fired. Scenario D
B22h defines a scenario assuming a predetermined conversation. For example, for a tour search, as shown in FIG. 15, an overseas wedding dialogue scenario, an overseas skiing dialogue scenario, and the like are prepared. Then, the scenario control unit 213 refers to the method definition DB 22i for the extracted action, and performs the action execution process (S
TEP1303). In this case, the scenario control unit 213
Transfers control to the search condition generation unit 214 or the search condition execution unit 215 depending on the content of the action to be executed. FIG. 16 is a diagram illustrating an example of the method definition DB 22i. For example, question_are to ask the area
In a, a question sentence "Where do you want to go?" is defined.

FIG. 17 is a flowchart for explaining the operation of the search condition generation unit 214 according to this embodiment. The search condition generation unit 214 performs a process of adding the content of the method obtained by the CGI interaction with the user to the search condition frame and registering the content in the user profile DB 22f. FIG. 18 is a diagram illustrating an example of the search condition frame. The search condition frame of this example is composed of a slot and a slot value, and the substance of the slot value is indicated by a pointer. The search condition frame preferably includes, for example, the transfer of a search condition by a concatenated character string of an argument group representing an individual search condition and the transfer of a search condition only by a reference pointer to a search condition storage object. However, the present invention is not limited to the type of intermediate data in which the data structures of the objects are temporarily stored.

The search condition generation unit 214 adds the content of the method given via the scenario control unit 213 as the value of the corresponding slot of the search condition frame (ST
EP 1701). Then, the search condition generating unit 214 stores the added search condition frame in the user profile DB 22
f (STEP 1702).

FIG. 19 is a flowchart for explaining the operation of the search condition execution unit 215 according to this embodiment. First, the search condition execution unit 215 checks the content DB2
2j is executed to generate a search sentence for searching (STE
P1901). This content DB search query generation process will be described later. The search sentence means a query (for example, an SQL sentence) for performing a search on the content DB 22j. The search condition execution unit 215 executes a search for the content DB 22j based on the generated search sentence (STEP 1902). Next, the search condition execution unit 215 checks the number of obtained items as a result of the search execution (STEP 1903), and if the obtained number is greater than 0, performs a score calculation process (STEP 190).
4) Sorting is performed in descending order of the score (STEP 190)
5). Then, the search condition execution unit 215 sets the return value to 0 and ends (STEP 1906).

On the other hand, if the number of contents acquired in STEP 1903 is 0, the search condition execution unit 215
Determines whether the search condition has been reset (S
TEP1907). If it is determined that the search condition has not been reset, the search condition is reset with reference to the specialized field dictionary DB 22d (STEP 1908). The search condition resetting process will be described later. Search condition execution unit 21
In step 5, after the search conditions are reset, the flow returns to STEP 1902 again. If it is determined that the search condition has been reset, the search condition execution unit 215 sets the return value to −1 and ends (STEP 1909).

FIG. 20 is a flowchart for explaining the search sentence generation processing for the content DB search shown in the above STEP1901. That is, the search query generation processing unit for content DB search is executed by the user profile DB 22f.
The value of the search item (slot) “area” is acquired with reference to the search condition frame stored in the “STEP 20”.
01). For example, according to the search condition frame shown in FIG. 18, “tie” is acquired as the value of “area”. Next, the search query generation processing unit for content DB search calls the specific keyword acquisition unit and performs a specific keyword acquisition process (STEP 2002). A concrete keyword (specific word) is a keyword that is semantically included in the abstract keyword (abstract word). In other words, if a certain keyword (for example, “Thailand”) is an abstract keyword,
Keywords semantically included in the keywords (for example, “Bangkok”, “Ayutthaya”, “Sukhothai”,...) Are the concrete keywords. The search query generation processing unit for content DB search sets the obtained embodied keyword (specific word) in the “area (accommodation place) specific word” of the search condition frame (STEP 2003). Next, the value of the search item “main purpose” in the search condition frame is obtained (STEP 20).
04) In order to further narrow down the search conditions for each of the elements “area”, “hotel”, “famous place / historic site” and “airline”, the materialized keyword acquiring unit is called, and the materialized keyword acquiring process is performed. Perform (STEP 2005-
2006). For example, for the purpose of "visiting archeological sites", when retrieving materialized keywords for "famous sites and historic sites", relics from the world such as "Sukhothai ruins" and "Great Wall" are acquired as reified keywords. Will be. As described later, these locations and a set of neighboring cities at these locations are stored in the specialized field dictionary DB2.
The search result is obtained using 2d, and the logical product with the concrete keyword of "Thailand" is calculated, and a check is performed so that logical contradiction does not occur as a search condition. In this example, “Ayutthaya” and “Sukhothai” are the logical product operation results. The search query generation processing unit for content DB search sets specific words in each of the corresponding elements of the search condition frame (STEP 2006). Next, the content DB
The search sentence generation processing unit for search refers to the specialized field dictionary DB 22d and performs a process of determining whether or not there is a logical inconsistency in the search condition (STEP2007). No), a search sentence for searching the content DB 22j is generated and stored in the user profile DB 22, and the return value is 0 (normal).
Is set, and the processing ends (STEP 2009). On the other hand, if it is determined that there is a logical contradiction, the search query generation processing unit for content DB search sets the return value to −1 and ends (STEP 2010).

FIG. 21 is a flowchart for explaining the details of the materialized keyword acquiring unit. The materialized keyword acquiring unit inputs the search source keyword (headword), the search target viewpoint, the search route viewpoint, the search route attribute, the departure date and the number of travel days as arguments, and outputs the materialized keyword with relevance as a return value. I do. That is, the specific keyword obtaining unit obtains a specific word for the search target viewpoint for the headword according to the condition given by the argument.
Here, the search target viewpoint is a viewpoint from which a specific word is to be acquired, and is a viewpoint intended for the scenario control unit 213 to acquire a search condition. For example, when the scenario control unit 213 attempts to acquire a specific condition from the viewpoint of “area” in the search condition, the viewpoint to be retrieved is “area”, and the embodied keyword acquiring unit determines the input area of “area”. An attempt is made to obtain a specific word of "area" from the headword. The search route viewpoint is a viewpoint to be searched which is set in advance for the search target viewpoint. Taking the search of tour information as an example, when acquiring the specific word “area” from the headword “I want to visit archeological sites” with the “theme” perspective, the headwords with the “theme” perspective have a hierarchical structure. The "theme" and "area" are set as search route viewpoints, assuming that they are constructed. The search route attribute is an attribute to be searched from a viewpoint to be searched (search route viewpoint). In this example, specific words and related words are set as search route attributes.

As shown in the figure, the materialized keyword obtaining unit determines whether or not a search target viewpoint exists in a headword (STEP 2101). If it is determined that there is a search target viewpoint in the headword, the materialized keyword acquisition unit performs a process of acquiring a search route viewpoint and a search route attribute fixed materialized keyword (STEP 2102), and ends the subroutine.

On the other hand, when it is determined that the search target viewpoint does not exist in the headword, the materialized keyword acquisition unit repeats the following processing for each of the search route viewpoints (STEPs 2103 to 2110). That is, the materialized keyword acquisition unit first determines whether or not the search route viewpoint exists in the headword (STEP 2103).
If it is determined that there is, it is further determined whether or not the current viewpoint (the search route viewpoint of interest) and the search target viewpoint are different (STEP 2104). When it is determined that the current viewpoint and the search target viewpoint are different, the materialized keyword acquiring unit performs a process of changing the current relevance (STEP 2).
105). In the relevancy changing process, for example, “current relevance = current relevance × relevance of the search route viewpoint ÷ maximum relevance” is calculated. Next, the specific keyword acquiring unit determines whether a specific word is specified in the search route attribute (STEP 2106), and if it is determined that the specific word is specified, performs a specific word search process (STEP 21).
07). Further, the materialized keyword acquiring unit determines whether a related word is specified in the search route attribute (ST).
EP2108), when judging that it is designated,
An area viewpoint is added to the search route viewpoint (STEP 210).
9), a related word search process is performed (STEP 211).
0).

FIG. 22 is a flow chart for explaining the operation of the retrieval route viewpoint / attribute fixed materialization keyword acquisition processing shown in the above STEP 2102. The search route viewpoint / attribute fixed materialized keyword acquisition unit acquires the same viewpoint as the search target viewpoint for the headword (STE
P2201). Next, the search route viewpoint / attribute fixed materialized keyword acquiring unit performs a process of changing the current relevance (STEP 2202). Currently, the reassociation process is
For example, this is performed by calculating “current relevance = current relevance × relevance of the search route viewpoint 観 点 maximum relevance”. Next, the search route viewpoint / attribute fixed materialized keyword acquisition unit determines whether or not a specific word exists in the viewpoint (STEP 2203). It is determined whether or not has already been output by the materialized keyword acquisition unit of the caller (STEP 2204). If the specific term has not been output yet, the search route viewpoint / attribute fixed concreted keyword acquiring unit outputs the specific word together with the relevance (S
TEP2205). On the other hand, when the search route viewpoint / attribute fixed concreted keyword acquisition unit determines that the specific word has already been output, the relevance of the specific word is further compared with the relevance of the same specific word already output. It is determined whether it is larger than (STEP 2206). When determining that the specific word has a higher relevance, the search root viewpoint / attribute fixed concreted keyword acquiring unit deletes the same specific word that has been already output together with the relevance (STEP 220).
7) Output the specific word together with the degree of relevance (STEP)
2205).

On the other hand, if it is determined in STEP 2203 that a specific word exists, the search route viewpoint / attribute fixed materialized keyword acquiring unit limits the search route viewpoint to the search target viewpoint, and further limits the search route attribute to the specific word. (Step 2208). Then, the search route viewpoint / attribute fixed materialized keyword obtaining unit performs a specific word search process (STEP 2209).

FIG. 23 is a flowchart for explaining the operation of the specific word search process. As shown in the figure,
The specific word search processing unit performs the following processing on all elements stored in an array of specific words (specific word collections serving as search starting points) in the viewpoint serving as a search source in the specialized field dictionary 22d (STEPs 2301 to 2304). ). That is,
The specific word search processing unit attempts to reach the keyword by following the link. By following the link, the current relevance is changed (STEP 2301). The replacement process of the current relevance is performed, for example, by calculating “current relevance = current relevance × current relevance / maximum relevance”. Next, the specific word search processing unit proceeds to STEP 2
By following the link in 301, a keyword is acquired (STEP 2302). In order to use the acquired keyword recursively as a headword, the (current)
It is replaced as a keyword (STEP 2303). Then, the specific word search processing unit recursively performs a process of acquiring a specific keyword (STEP 2304).

FIG. 24 is a flowchart for explaining the operation of the related word search process. As shown in the figure,
The related word search processing unit performs the following processing for all elements stored in an array of related words (related word collection serving as a search starting point) in a viewpoint serving as a search source in the specialized field dictionary 22d (STEPs 2401 to 2405). ). That is,
First, the related word search processing section performs a process of acquiring the degree of relevance this time (STEP 2401). The relevance this time is obtained by giving the departure date and the number of travel days. Next, the related word search processing unit attempts to reach the keyword by following the link. By following the link, the current relevance is changed (STEP 2402). The process of changing the current relevance is, for example, “current relevance = current relevance ×
This is performed by calculating “relevance this time ÷ maximum relevance”. Next, the related word search processing unit obtains a keyword by following the link in STEP 2401 (STEP 2403), and replaces the obtained keyword as the (current) keyword in order to use the obtained keyword recursively as a headword ( (STEP 2404). Then, the related word search processing unit recursively calls the materialized keyword processing unit to perform a materialized keyword acquisition process (S
TEP2405).

FIG. 25 and FIG.
It is a flowchart for demonstrating operation | movement of the search condition logic contradiction determination processing shown by 07. That is, in FIG. 25, the search condition logic contradiction determination processing unit determines whether or not an area (accommodation place) specific word and a main destination area (accommodation place) specific word exist in the search condition frame (STEP 2).
501 to 2503). If an area (accommodation place) specific word exists and a main destination area (accommodation place) specific word exists, the search condition logical contradiction determination processing unit
The logical product of the specific word and the main target area (accommodation place) specific word is calculated (STEP2504), and it is further determined whether or not the element of this logical product exists (STEP2505). If it is determined that there is a logical product element, the search condition logical contradiction determination processing unit determines the logical product as a search condition for the accommodation place (STEP 2506). If there is no element of the logical product, the return value is set to −1 as a logical contradiction (STEP 2512), and the processing ends.

On the other hand, if the specific area (accommodation place) word does not exist and the specific main target area (accommodation place) specific word exists, the search condition logical contradiction determination processing unit determines whether the main target area (accommodation place) specific word is present. Words as search conditions for accommodation (ST
EP 2507). In addition, if there is an area (accommodation place) concrete word and there is no main target area (accommodation place) concrete word,
The search condition logical contradiction determination processing unit determines the specific area (accommodation place) word as the accommodation condition search condition (STEP 250).
8). In addition, when there is no area (accommodation place) specific word and there is no main destination area (accommodation place) specific word,
The search condition logical contradiction determination processing unit sets this as a logical contradiction and sets the return value to −1 (STEP 2512).

Next, the search condition logic contradiction determination processing section determines whether or not the element of the main target hotel concrete word exists (S).
(TEP2509), when it is determined that there is an element, it is determined whether or not an element exists in the logical product of the location of the main target hotel concrete word and the determined accommodation place (STEP 2510). This is realized by linking, on the specialized field dictionary 22d, a keyword having an area perspective meaning its location to a related word of the hotel perspective of the keyword representing the hotel name, which is an element of the main target hotel concrete word. You. If it is determined that there is an element in the logical product, the logical product is added to the accommodation,
In addition, among the elements of the main target hotel specific word, the hotel having the logical product at the address is determined as a hotel search condition (STEP 2511).

Next, the search condition logic contradiction determination processing section determines whether or not the main destination famous place / historic concrete word element exists (STEP 2513). If it is determined that the element exists, the specialized field dictionary is used. With reference to 22d, the neighborhood city (related word) of the accommodation place determined as the search condition is acquired, and the accommodation place determined as the search condition is also added to form a neighborhood city set (STEP 2514). Next, the search condition logic contradiction determination processing unit determines whether or not there is an element of a logical product of the location of the main destination famous place / historic concrete word and the set of neighboring cities (ST).
If it is determined that the logical product exists, the main target famous place / historical site specific word having the logical product is determined as a search condition of the famous place / historical site (STEP 2516). Further, the search condition logic contradiction determination processing unit determines whether or not the element of the main target airline specific word exists (STEP 2).
517) If it is determined that the airline exists, the main target airline specific word is determined as the airline of the search condition (ST).
(EP2518), the return value is set to 0 as normal termination (STEP2519), and the processing ends. In addition, STEP2
If it is determined in 515 that there is no logical product, the search condition logical contradiction determination processing unit determines that there is a logical contradiction, sets -1 as a return value, and ends the processing. (STEP
2520).

FIG. 27 is a flowchart for explaining the operation of the score calculation processing shown in STEP 1904. The score calculation processing unit declares data for storing the score and initializes the score for each content to 0 (S
TEP2701). Until there is no result set of the acquired content, the score calculation processing unit sequentially orders the elements of “accommodation place”, “hotel”, “famous place / historic site”, and “airline” (herein, referred to as “search items”). The following processing is repeated (STEPs 2702 to 2703). That is,
The score calculation processing unit determines whether or not an element exists in the search item of the search condition frame (STEP 270).
2) If it is determined that the search item exists, the same data as the search item in the content is selected from the search item in the search condition frame, and the product of the relevance and the weighting frequency of the search item is stored in the score. Is added to the data for use. For example, the score of the content ID “TRA01213133” is that the accommodation place is “Bangkok” and the famous place / historic site is “Ayutthaya archeological site”.
95 × 1 + 88 × 1.5 = 227.

FIG. 28 is a flowchart for explaining the operation of the search condition resetting processing section shown in STEP 1908. First, the search condition reset processing unit copies the original search condition frame, and updates the value of the search item “area” of the copied frame to null (STEP 2).
801). Next, the search condition resetting processing unit refers to the specialized field dictionary 22d and performs a process of determining a logical inconsistency in the search condition for resetting (STEP2802). As a result of the logical inconsistency determination processing, when it is determined that the logical inconsistency (return value is -1) (YES in STEP 2803), the return value is set to -1, and the processing is ended (STEP 2804). On the other hand, when determining that there is no logical contradiction,
Referring to the user profile DB 22f, a search sentence is generated and stored, and then the return value is set to 0 and the processing is terminated (STEP 2805).

FIG. 29 and FIG.
13 is a flowchart for explaining the operation of the search condition logical contradiction determination process for resetting shown in FIG. That is, in FIG. 29, the search condition logical contradiction determination processing unit for resetting determines whether or not the main target area (accommodation place) specific word and the main target hotel specific word of the search condition frame exist (STEPs 2901 to 2903). ). If the main target area (accommodation place) specific word exists and the main target hotel specific word exists, the search condition logical contradiction determination processing unit for resetting uses the main target area (accommodation place) specific word and the main target hotel. The logical product with the location of the specific word is calculated (STEP 2904), and it is further determined whether or not the element of this logical product exists (STE).
P2905). If it is determined that there is a logical product element, the resetting search condition logical contradiction determination processing section determines the logical product as a search condition of the accommodation place (STEP 290).
6). If there is no element of the logical product, the return value is set to −1 as a logical contradiction (STEP 2922),
finish.

On the other hand, if the main target area (accommodation place) specific word does not exist and the main target hotel specific word exists, the search condition logical contradiction determination processing unit for resetting determines the location of the main target hotel specific word. Is determined as a search condition for accommodation (ST
EP 2907). In addition, when the main target area (accommodation place) specific word exists and the main target hotel concrete word does not exist, the search condition logical contradiction determination processing unit for resetting sets the main target area (accommodation place) specific word to the accommodation place. Confirm as search condition (ST
EP 2908). If the main target area (accommodation place) specific word does not exist and the main target hotel specific word does not exist, the search condition logical contradiction determination processing unit for resetting is executed in STEP.
It moves to 2913.

Next, the search condition logical contradiction determination processing unit for resetting determines whether or not the element of the main destination famous place / historical concrete word exists (STEP 2909 in FIG. 30). Then, referring to the specialized field dictionary DB 22d, a neighborhood city which is a related word of the accommodation place determined as the search condition is obtained, and the accommodation place determined as the search condition is added to form a neighborhood city set (STEP 2910). The retrieving search condition logical contradiction determination processing unit determines whether or not there is an element of a logical product of the location of the main destination famous place / historic concrete word and the set of neighboring cities (STEP 2911). Then, the main target landmark / historical site specific word having the logical product at the address is determined as a search condition of the landmark / historic site (STEP 29).
12). In addition, STEP2909 and STEP291
If there is no logical product element in 1, the return value is set to -1 as a logical contradiction, and the processing ends (STEP 29).
21).

On the other hand, as described above, when there is no specific word for the main destination area (accommodation place) (No in STEP 2901), and there is no specific word for the main destination hotel (STEP 2).
901: No), it is determined whether or not the element of the main destination famous place / historic concrete word exists (STEP 2913 in FIG. 30).
The search condition logical contradiction judgment processing unit for resetting
Attempt to acquire accommodation from the location of the historic site (STEP 291
4) If it can be obtained (YES in STEP 2915)
s), the acquired accommodation place is determined as a search condition accommodation place (STEP 2916), and the main target attraction / historical site concrete word is determined as a attraction / historical location search condition (STEP 291).
7). In STEP 2913, the main destination
If there is no element of the historical concrete word, and STEP2
If it is determined in 915 that the accommodation place cannot be obtained from the location of the famous place or historic site, the flow shifts to STEP2918. Also,
In STEP 2914, a keyword having an area viewpoint meaning its location is linked on the specialized field dictionary 22d to a keyword related to a sightseeing point / historical point of view, which is a keyword representing the name of a sightseeing point / historic point, which is an element of the main purpose / point of interest. Further, when a keyword having another area viewpoint is linked to a related word in the area viewpoint of the keyword, this is realized by determining the latter as an accommodation place.

When the search condition of the famous place / historical site is determined by the above processing, the search condition logical contradiction determination processing unit for resetting,
It is determined whether or not the element of the main target airline specific word exists (STEP 2918). If it is determined that the element exists, the main target airline specific word is determined as the airline of the search condition, and the return value is set to 0. Set to (STEP 29
20), end.

FIG. 31 is a diagram showing an example of a screen presented by the information providing system according to the present embodiment. That is,
The figure shows an example of a screen in which a user is interactively inputting search conditions in a natural language format with an information providing system. The information providing system includes a scenario control unit 2
13, while presenting the method in the system response area 311, the user inputs search conditions in a natural language format in the user input area 312, and selects the “OK” button. Sent to the information providing system. In addition, the information providing system displays a dialog history between the user and the information providing system in the dialog history area 313. As described above, the information providing system prompts the user to input necessary search conditions under the control of the scenario control unit 213. Then, when the user selects the “end” button at a stage where the search condition has been input to some extent, the search condition execution unit 215 executes the search, and presents the search result shown in FIG. 32 to the user, for example.

[Second Embodiment] This embodiment is a modification of the above-described embodiment, in which a search for a database in which information is stored is advanced by a natural sentence dialogue with a user, and It is characterized in that while providing search results, information about the user is provided to the client company that supplied the information.

For example, taking the information providing system for searching and providing tour information (tour products) shown in the above embodiment as an example, the user can use the information providing system to
When the tour information is searched by inputting a fragmentary natural sentence character string about the travel plan desired by each, the user information is sent to a travel agency that can provide the tour product indicated as a search result. (Customer log). In this case, it is desirable to provide only when it is desired to transmit a request to a travel company selected by the user from travel companies that can be provided. The user information includes, for example, the mail address of the user, the content of the input history at the time of search, and the like. The travel agency can advantageously promote sales of tour products to the user based on the provided customer transfer information. When a travel agency performs a sales activity with a user via information provision of the information providing system, the travel company feeds back the result to the information providing system. The information providing system accumulates as contract information based on the feedback result, and uses it for the next information provision and product planning.

FIG. 33 is a block diagram showing a functional configuration of the information providing system according to the present embodiment. The information providing system according to the present embodiment includes a search engine unit 2
1 ', a first database unit 22 (corresponding to the database unit 22 in the above embodiment), a contract information management unit 23, and a second database unit 24.

The search engine unit 21 'further includes a travel company information generation unit 216. Travel agency information generator 2
16 acquires and generates information (travel company information) about the travel company related to the search result of the user from the travel company DB 24a.

The contract information management unit 23 includes a customer log output unit 2
31, customer sending data output unit 232, contract data creation unit 23
3 and a contract data statistical processing unit 234. The customer log output unit 232 allows the user who has completed the search by performing a series of interactive inputs to provide the travel agency indicated in the final search result with further information for applying for a tour product. For example, when user information such as an e-mail address is submitted as desired, customer information (customer log) to be provided to the travel agency is generated and transmitted to the client computer 2B of the travel agency.

Upon receiving the request for the contract information registration form from the client computer 2B of the travel agency, the customer data output unit 232 extracts a record matching the predetermined condition from the records registered in the customer log DB 24c. Is transmitted to the client computer 2B. The predetermined condition is arbitrarily set as a weight in the travel agency DB 24a according to a customer characteristic (user characteristic) to be emphasized. For example, when it is desired to extract records of a user with a high contract unit price, a user with a high contract rate, and a user with a short contract day as a past record, the corresponding weight value is set large. The contract unit price, the contract rate, the contract days, and the like are registered in the contract user DB 24d, and are sequentially updated by feedback of the contract information from the travel agency.

The contract data creation unit 233 transmits the contract information registration form to the client computer 2B.
When a registration request for contract information is received from, the contract data is created in accordance with the parameters included in the registration request and registered in the contract user DB 24d. The contract data statistical processing unit 234 performs predetermined statistical processing and the like based on the created contract data, and outputs the processing result to the contract user D.
Register in B24d.

The second database 24 is a travel agency DB
24a, category-specific travel agency DB 24b, customer log DB
24c and a contract user DB 24d. The travel company DB by category 24b is a travel agency DB by main purpose.
24b1, a destination-specific travel company DB 24b2, and a hotel-specific travel company DB 24b3, and the actual results (number of contracts) of the travel companies in each category are registered.

FIG. 34 is a flowchart for explaining the operation of the search condition execution unit 215 'according to the present embodiment. The search condition execution unit 215 'is basically the same as the process of the search condition execution unit 215 shown in FIG.
The difference is that a travel company information generation process of STEP 3406 and STEP 3410 is added. Although not shown, the search condition execution unit 215 ′
A search ID is issued before the first content DB search query generation process. This search ID is used for unique management by giving relevance to each process described later.

FIG. 35 is a flow chart for explaining the operation of the travel company information generation processing section 216 according to the present embodiment. As shown in FIG.
6 first refers to the user profile DB 22f,
The contents of a search condition frame of a user who is interactively performing a search are acquired (STEP 3501). The travel company information generation unit 216 checks whether or not the number of travel company display flags is one or more (STEP 3502), and if it is zero, performs travel company information acquisition processing (STEP 3).
503). On the other hand, if the number of travel agency display flags is one or more, the number of acquired content is further checked (STEP 3504). If the number is zero, travel agency information acquisition processing is performed (STEP 3503).

FIG. 36 is a flowchart for explaining the travel company information acquisition processing shown in STEP 3503. As shown in the figure, the travel company information generation unit 216
The following processing is performed on each element of “main purpose”, “fixed accommodation place”, and “fixed hotel” in the search condition frame (S
TEP 3601-603). That is, the travel company information generation unit 216 refers to the user profile DB 22f, acquires the corresponding item in the search condition frame (STEP 3601), and checks whether the content of the item is Null (STEP 3602). The travel agency information generation unit 216 determines that the content of the acquired item is Nu.
If it is determined that the item is not “11”, the contents of the item are used as headwords, and the category-specific travel company DBs 24b1 to 24b3
Are searched, and the number of contracts for each travel agency code is added to the travel agency code file (STEP 360).
3). More specifically, the travel company information generation unit 216
When "diving" is set in "main purpose" of the search condition frame, the main purpose heading word "diving" is referred to by referring to the travel company DB 24b1 by main purpose as a heading word. Get the number of contracts.
Next, when "Phi Phi Island" is set in "Confirmed accommodation place", this is used as a headword to make travel destination DB by destination.
With reference to 24b2, the number of contracts made for the travel agency whose destination headword is “Phi Phi Island” is acquired. At this time, if the travel agency is the same, the number acquired this time is added to the number of contracts already acquired. Similarly, for "confirmed hotels",
Obtain and add the number of closed deals.

Next, the travel company information generation unit 216 sorts the records of the travel company code file created in this manner in descending order using the number of contracts as a key (S
TEP3604), travel agency DB according to the contents
24a is updated (STEP 3601). As a result, a travel agency having a track record for each element can be obtained according to the input search conditions.

FIGS. 37 to 39 show travel companies D by category.
It is a figure which shows the data structure of B24b1-24b3, More specifically, FIG. 37 is travel company DB24b1 classified by main purpose.
FIG. 38 is a diagram showing a data structure of a destination, and FIG.
FIG. 39 is a diagram showing a data structure of a travel company DB 24b3 by hotel. The category-specific travel company DB 24b is a collection of categories that each travel company excels at. For each headline in each category (for example, “main purpose”, “destination”, and “hotel”), a contract is made for each travel company. The number is kept.

FIG. 40 is a diagram showing a data structure of the travel agency DB 24a. Information about the travel company is registered in the travel company DB 24a in advance. That is, as shown in the figure, the travel agency DB 24a stores “travel agency code”,
"Travel company name", "Travel company URL", "Email address", "Phone number", "Margin", "Margin classification", "Weighted deals", "Average closed rate weighted", "Average closed days weighted" And “average unit weight” fields. In the “weighting of the number of deals”, “weighting of the average closing rate”, “weighting of the average number of closing days”, and “average unit price weighting”, weight values of the customer characteristics that are emphasized by the travel agency are set. For example,
When a certain travel company attaches importance to a user with a large number of contracts, the weight value is set higher than that of another travel company.

A user who has completed a search by performing a series of interactive inputs submits, for example, an e-mail address in order to request further information from the travel agency indicated in the final search result. , The customer log output unit 231
A customer log is generated, registered in the customer DB 24c, and sent to the corresponding travel agency. For example, FIG.
It is assumed that a search result screen as shown in FIG. In the figure, three tour products that match the search conditions input by the user are presented. If the user wishes to provide further information on this search result screen, he / she checks the check box 411 of the desired tour product, inputs an e-mail address in the e-mail address field 412, and selects the send button 413. I do. In response to this, the information providing system generates customer sending data in the customer sending data output unit 232 and sends it to the corresponding travel agency.

FIG. 42 is a flowchart for explaining the operation of the customer log output unit 231 according to this embodiment. As shown in the figure, the customer log output unit 231 performs the following STEP 420 for the travel company selected by the user.
1 to 4204 are performed. That is, the customer log output unit 231 refers to the user profile DB 22f, and acquires, from the search condition frame, items corresponding to the search condition input contents and items corresponding to the search condition finalized contents, respectively (STEP 4201 and 4202). Next, the customer log output unit 231 generates a record based on these acquired items and adds the record to the customer log DB 24c (ST
EP4203). FIG. 43 is a diagram illustrating a data structure of the customer log DB. The customer log output unit 231 is a travel agent D
With reference to B24a, the contents of each record in the customer log DB 24c are transmitted to the corresponding travel agency (S
TEP4204). In the present embodiment, since both the search condition input content and the confirmed search content are sent to the travel agency, the travel agency that receives the information grasps what kind of search the user performs and what the user wants. Will be able to do that. Then, when the above processing is completed for each travel company selected by the user, the customer log output unit 231 adds 1 to the number of customer search IDs of the user in the contracted user DB 24d, and ends (STEP 4).
205).

As described above, the information providing system sends the customer log to the travel agency, presents the customer delivery data in response to the request from the travel agency, and receives feedback on the user's contract status. . That is, when the information providing system receives a request for a contract information registration form from a travel agency, the customer data output unit 233 creates a contract information registration form and sends it to the travel agency. The customer sending data output unit 233 extracts a record to be presented according to the non-display flag of the customer sending log DB 24c and the past closing information registered in the closing user DB 24d, creates a closing information registration form, and sends the form to the travel agency. Send out. The operation of the customer data output unit 233 will be described later in detail.

FIG. 44 is a diagram showing an example of the contract information registration form screen. As shown in the figure, on the contract information registration form screen, the operator can confirm whether or not the tour product requested by the user has been closed, the amount of the tour product, and the search result at the time of the next search by the user. Can be input as to whether or not to display. When the operator selects the registration button, the entered content is made as the contract data, ie, the contract data creation unit 233.
Will be accepted by

FIG. 45 is a flow chart for explaining the operation of contract data creation section 233. The contract data creation unit 233 performs the following processing on all records included in the transmitted contract data. That is, the contract data creation unit 233 checks whether or not the contract flag is set for the record managed by the first search ID in the contract data transmitted from the travel agency (STEP 4501). When determining that the deal flag is set, the deal data creation unit 233 updates the content of the customer log DB 24c according to the travel company code and the search ID (STEP 450).
2), a contracted user DB is created (STEP 45)
03). On the other hand, if the contract flag is not set to 1, the contract data creation unit 233 sets the non-display flag to 1
It is checked whether it is set to (STEP4)
504). When determining that the non-display flag is set to 1, the contract data creation unit 233 updates the non-display flag of the corresponding record in the customer log DB 24c to 1 according to the travel agency code and the search ID. (STE
P4505).

Note that in STEP 4502, the above-described ST
Travel company DB24b by main purpose described in EP3603
1. When storing the number of contracts in the destination-specific travel company DB 24b2 and the hotel-specific travel company DB 24b3, the configuration may be such that the flag is used to refer to which of the main purpose, the destination, and the hotel is taken into account. Good.

FIG. 46 shows a contracted user D according to the present embodiment.
13 is a flowchart for explaining a B creation process. Upon accepting the contract information registration form, the contract data creation unit 233 adds 1 to the number of contracts of the user managed by the user ID, calculates an average contract rate, and updates the record (STEP4601). The average contract rate is calculated by, for example, average contract rate = number of contracts / number of customer search IDs. Next, the contract data creation unit 233 calculates the average number of contract days for the user and updates the record (STEP4602). The average number of contract days is calculated, for example, as follows: average contract days = (current contract days + previous average contract days × previous search ID number) / current customer search ID number. Here, the number of days closed this time is the number of days from when the customer log is created to when the current contract is received, and is calculated as follows: days closed this time = date closed this time-date transferred this time. Then, the contract data creation unit 233 calculates the contract unit price of the user and updates the record (S
TEP4603). The contract price is calculated, for example, as follows: contract price = (current contract price + previous contract price × previous contract number) / current contract price.

FIG. 47 is a flowchart for explaining the operation of the contract data statistical processing section 233 according to the present embodiment. The contract data statistical processing unit 233 is executed when the contract data is created by the contract data creating unit 233.

As shown in the figure, the contracted data statistical processing section 233 refers to the contracted user DB 24d and extracts records in which the number of contracts is 1 or more (STEP 4701).
The average value and standard deviation of each of the attributes (the number of contracts, the average closing rate, the average number of closing days, and the unit price of the contract) in the extracted record are calculated (STEP 470).
2). Then, the contract data statistical processing unit 233 calculates deviation values for each of the number of contracts, the average contract rate, the average number of contract days, and the contract unit price for all the records in which the number of contracts is one or more, and updates the record. (Step 4703). The deviation value for the number of contracts, the average contract rate, and the unit price of the contract are calculated, for example, as follows: deviation value = (difference from the average) × 10 / standard deviation + 50. Is calculated such that, for example, deviation value = − (difference from the average) × 10 / standard deviation + 50. FIG. 48 is a diagram showing an example of the data structure of the contracted user DB created as described above.

FIG. 49 is a flowchart for explaining the operation of the customer sending data output unit 232 according to the present embodiment. As shown in the figure, upon receiving a request for a contract information registration form from a travel company, the customer data output unit 232, for each user whose travel company non-display flag is 0 in the customer log DB 24c, closes the contract user While acquiring each deviation value of DB24d (ST
(EP4901), referring to the travel agency DB 24a to acquire weighting data of the travel agency (STEP 490).
2). Next, the customer sending data output unit calculates scores for all user records (STEP 490).
3). The score is calculated by, for example, score = Σdeviation × weight. After calculating the scores for the records of all the users, the customer sending data output unit 233 sorts the records in descending order using the calculated scores as keys (ST).
EP 4904). Then, the customer sending data output unit 233
According to the user ID of the sorted records, a record in which the non-display flag of the travel agent log DB of the corresponding travel agency is 0 and the deal flag is 0 is acquired from the customer log DB 24c and output as a deal information registration form. (STE
P4905).

As described above, according to the present embodiment, when the user takes an action to provide more information with respect to the final search result presented by the interactive input with the user, The user information including the input history at the time of the search is transmitted as a customer log to the client company that provided the information. As a result, the client company that has received the customer transfer log can proceed with effective sales activities for the user. Further, according to the present embodiment, the client company is managed as the contract information by feeding back the result of the sales activity based on the customer sending log to the information providing system. Since the client company can refer to the managed contract information at any time, effective sales activities can be promoted.

Further, the interactive input with the user is a fragmentary element of the travel plan desired by each user,
If the client company is a travel agency, it has the following advantages.

That is, for the user, a phrase relating to a fragmentary travel plan (for example, “a beautiful place in the sea”)
, Specific words (for example, “Phi Phi Island”,...) Sufficient for performing a conditional search for package travel information are obtained, and a search result can be obtained. In the search results, not only tour information but also travel company information can be browsed separately as search results, so even if there is no desired tour, advertising expenses are not always actively invested in promotion activities for packaged travel planning, but appropriate You can browse travel agencies (for example, specializing in tours to Phi Phi Island) that are unique in their prices and services. In addition, it is possible to browse along with the travel company's comment information (such as "If you dive," etc.) from the travel companies with the highest track record of past contracts. Inquiry), and efficient and satisfactory travel is encouraged.

Further, the travel agency can not only obtain information necessary for contacting a prospective customer (a user who has transmitted a request), but also obtain information in accordance with arbitrary pre-registered information (including weighting relating to user characteristics). The customer transfer log can be referred to, and the effective sales activity can be advanced for the prospective customer sent by using the contract information registration form as needed.
For example, a travel agency that has registered package tour information,
In some cases, it is not possible to respond to the user's request sent to the customer because the tour is unavoidably canceled because the number of package tour products does not reach the lowest. In such a case, the travel agency can refer to the search condition input content and the fragmentary travel plan phrase of the fixed search content included in the customer log,
Rather than simply selecting and presenting products that resemble tour products that have been canceled, present alternatives that are more supported by the basic image of the fragmentary travel plan that users are seeking for products. Can be.

Further, even a travel company that does not register package tour information, for example, a travel company that is good at selling custom tour products, can refer to the contents included in the customer log to confirm necessary items with the user. Instead of spending time, services such as presenting custom tour products that meet the needs, such as notifying the user of more detailed characteristics, can be provided.

More specifically, since words related to a fragmentary travel plan of a potential user at that time can be read from the customer log, the purpose, destination, hotel, etc., of the headline can be determined through tabulation analysis and the like. Since it can be determined whether or not there is a response, the travel company can effectively perform product planning and notification including a travel viewpoint (travel condition) that the travel company is good at.

Further, it is not necessary to create a business process list for each prospective customer's inquiry by using the contract information registration form, and it is possible to manage whether contracts have been completed and whether or not to display them in the list next time. Can be significantly increased. In the case of management for each inquiry matter, it is possible to eliminate a mistake that is likely to occur in the list management for each user and, when the same user makes a plurality of inquiries with different travel dates, the other is left unsupported.

Further, for example, for a travel agency that wants to run a tour with less than the number of people running and that is close to the date and time of the tour, there may be cases where it is more desirable to preferentially deal with a user who has made a deal and made a deal without taking much time. is there. Conventionally, it was cumbersome when individual management such as the number of past contracts of the user for each inquiry matter, the rate at which the contract was completed, the number of days until the contract was completed, the unit price of the travel cost that was concluded, but it was complicated, but the weight value was set in advance. By doing so, they can be browsed in order, and customer service becomes naturally effective. It is possible to estimate the sales prospect even at the time of sending a customer who does not reach a contract. In addition, a travel agency that wants to take measures in accordance with the business plan can achieve plan optimization by changing the registration information from time to time as control data set information such that each weight changes from time to time, along with estimation estimates (weekend) Users with long days to close at the last applicable date and time, users with low actual unit price at the end of the month after payday, and high actual unit price and closing rate if the company's target sales are expected to be less than the end of the term It is also very useful because it is possible that the user characteristics and the business processing date and time that are expected to be correlated, such as a user having a high degree of influence, are expected to affect the business plan.

The above embodiment is an exemplification for explaining the present invention, and is not intended to limit the present invention to only these embodiments. The present invention, unless departing from the gist thereof,
It can be implemented in various forms. For example, the operation of the function realizing means has been described sequentially, but the present invention is not particularly limited to this. Therefore, as long as no inconsistency occurs in the operation, the order of processing may be changed or the operation may be performed in parallel.

[0083]

According to the present invention, an interactive information providing system based on natural sentences is provided.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a schematic configuration of an information providing system according to an embodiment of the present invention.

FIG. 2 is a block diagram showing a functional configuration of an information providing system according to an embodiment of the present invention.

FIG. 3 is a block diagram for explaining the operation of a natural sentence analysis unit.

FIG. 4 is a diagram illustrating a syntax analysis process.

FIG. 5 is a diagram illustrating a syntax analysis process.

FIG. 6 is a diagram illustrating a syntax analysis process.

FIG. 7 is a flowchart illustrating the operation of a dialog control unit.

FIG. 8 is a diagram illustrating an example of an emotion dictionary DB.

FIG. 9 is a diagram illustrating an example of a logical expression DB.

FIG. 10 is a diagram illustrating an example of an expression conversion DB.

FIG. 11 is a diagram illustrating a conceptual configuration of a specialized field dictionary DB.

FIG. 12 is a diagram for explaining a data structure of a specialized field dictionary DB.

FIG. 13 is a flowchart illustrating the operation of the scenario control unit.

FIG. 14 is a diagram illustrating an example of an action rule base DB.

FIG. 15 is a diagram illustrating an example of a scenario DB.

FIG. 16 is a diagram illustrating an example of a method definition DB.

FIG. 17 is a flowchart illustrating an operation of a search condition generation unit.

FIG. 18 is a diagram illustrating an example of a search condition frame.

FIG. 19 is a flowchart illustrating the operation of a search condition execution unit.

FIG. 20 is a flowchart illustrating a search query generation process for a content DB search.

FIG. 21 is a flowchart illustrating details of a specific keyword acquisition process.

FIG. 22 is a flowchart illustrating an operation of a search route viewpoint / attribute fixed materialized keyword acquisition process.

FIG. 23 is a flowchart illustrating the operation of a specific word search process.

FIG. 24 is a flowchart illustrating an operation of a related word acquisition process.

FIG. 25 is a flowchart illustrating an operation of a search condition logical contradiction determination process.

FIG. 26 is a flowchart illustrating the operation of a search condition logical contradiction determination process.

FIG. 27 is a flowchart illustrating an operation of a score calculation process.

FIG. 28 is a flowchart illustrating an operation of a search condition resetting process.

FIG. 29 is a flowchart illustrating the operation of a search condition logical contradiction determination process for resetting.

FIG. 30 is a flowchart illustrating the operation of a search condition logical contradiction determination process for resetting.

FIG. 31 is a diagram illustrating an example of an input screen presented by the information providing system.

FIG. 32 is a diagram illustrating an example of a search result screen presented by the information providing system.

FIG. 33 is a block diagram illustrating a functional configuration of an information providing system according to an embodiment of the present invention.

FIG. 34 is a flowchart illustrating an operation of a search condition execution unit.

FIG. 35 is a flowchart illustrating the operation of a travel company information generation processing unit.

FIG. 36 is a flowchart illustrating a travel company information acquisition process.

FIG. 37 is a diagram showing a data structure of a travel agency DB for each main purpose.

FIG. 38 is a diagram showing a data structure of a destination-specific travel company DB.

FIG. 39 is a diagram showing a data structure of a travel agency DB by hotel.

FIG. 40 is a diagram showing a data structure of a travel agency DB.

FIG. 41 is a diagram showing an example of a search result screen presented by the information providing system.

FIG. 42 is a flowchart illustrating the operation of a customer log output unit.

FIG. 43 is a diagram showing a data structure of a customer sending log DB.

FIG. 44 is a diagram showing an example of a contract information registration form screen presented by the information providing system.

FIG. 45 is a flowchart for explaining the operation of a contract data creation unit.

FIG. 46 is a flowchart illustrating a contracted user DB creation process.

FIG. 47 is a flowchart for explaining the operation of a contract data statistical processing unit;

FIG. 48 is a diagram showing an example of a data structure of a contract user DB.

FIG. 49 is a flowchart for explaining the operation of a customer data output unit.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Server computer 2 ... Client computer 11 ... Web server program 12 ... Web page 13 ... CGI program 14 ... Database 15 ... Browser 21 ... Search engine unit 22 ... Database unit 211 ... Natural sentence analysis unit 212 ... Dialogue control unit 213 ... Scenario Control unit 214: search condition generation unit 215: search condition execution unit

Claims (18)

[Claims] 1. A user is urged to input a character string in a natural sentence format, a predetermined natural language process is performed on the character string input by the user to obtain a predetermined keyword, and the obtained predetermined keyword is obtained. Creating a search condition frame having predetermined search items based on the search condition, searching the first database based on the created search condition frame, and presenting the search result to the user. Information provision method. 2. The information providing method according to claim 1, wherein a predetermined scenario is stored, a character string in a natural sentence format is presented to the user according to the predetermined scenario, and the user is prompted to input. Method. 3. The method according to claim 1, wherein the keyword obtained by the predetermined natural language processing is converted into a keyword in a basic format such that its fluctuation expression is absorbed.
Or the information providing method described in 2. 4. A reference is made to a second database that stores keywords that are classified for each predetermined viewpoint and that are associated with each other, and keywords that are semantically embodied from the acquired keywords are specified. 4. The information providing method according to claim 1, wherein a search condition frame is created based on a keyword. 5. A keyword classified into a viewpoint different from a viewpoint of the keyword is specified from the acquired keywords by referring to a second database storing keywords associated with each other and associated with each other in a predetermined viewpoint. 5. The information providing method according to claim 1, wherein a search condition frame is created based on the specified keyword. 6. A keyword obtained by giving a weight value indicating a degree of relevance to the keywords to be associated with each other, and searching a first database based on the created search condition frame according to the weight value. The information providing method according to claim 4, wherein each score of the search result is calculated, and the search result is presented in the order of the calculated score. 7. The information providing method according to claim 4, wherein a character string in a natural sentence format is presented so as to prompt the user to enter a keyword to be classified for each of the predetermined viewpoints. 8. The information providing method according to claim 1, wherein when a natural sentence indicating the end is input from the user, the subsequent processing is interrupted. 9. Searching the first database, if the first search result is 0, changes the content of a specific search item in the search condition frame to null;
9. The information providing method according to claim 1, wherein a re-search is performed. 10. The information provision according to claim 1, further comprising: prompting a user to select the search result, and providing information relating to the user as customer transfer information to a company associated with the selected search result. Method. 11. The information providing method according to claim 10, wherein the customer sending information includes the input character string and a search condition character string determined based on the character string. 12. A recording medium on which a program for causing a server computer to realize a predetermined function is recorded, wherein the program prompts a user on a client computer to input a character string in a natural sentence format. A function of performing a predetermined natural language process on a character string input by a user to obtain a predetermined keyword, a function of creating a search condition frame having a predetermined search item based on the obtained predetermined keyword, And a function of searching the first database based on the created search condition frame and presenting the search result to the user. 13. A means for prompting a user to input a character string in a natural sentence format; a means for performing a predetermined natural language process on a character string input by the user to obtain a predetermined keyword; Means for generating a search condition frame having a predetermined search item based on the predetermined keyword, and searching the first database based on the generated search condition frame, and presenting the search result to the user And an information providing device. 14. Tour information for accepting a search condition from a user in an interactive manner using a natural sentence, searching a content database storing information about the tour according to the search condition, and providing the search result to the user as information about the tour. A providing system, comprising: a natural language database storing natural language processing rules; a first viewpoint relating to the purpose of the tour; and a second viewpoint relating to the destination of the tour, and keywords classified for each viewpoint. A specialized field database that stores the groups in association with each other, a natural language processing unit that refers to the natural language database, performs natural language processing on a natural sentence given by the user, and acquires a predetermined keyword Refers to the specialty database and is entered by the user A keyword acquisition unit configured to acquire a keyword that is semantically embodied from the acquired keywords so that a tour that matches the search condition is embodied; and a predetermined search item based on the embodied keyword. Search condition frame creation means for creating a search condition frame having the same; search execution means for searching the content database based on the created search condition frame; presentation for presenting a search result obtained by the search execution means to a user Means for providing tour information. 15. A step of generating customer transfer data based on contract data stored in a contract user database in response to a request from a client computer, and a step of presenting the generated customer transfer data to the predetermined company. Receiving the contract information for each user for the presented customer transfer data, and updating the contract data stored in the contract user database based on the received contract information. How to provide information. 16. The step of generating the customer transfer data includes: calculating a score for each user according to an evaluation value given to a predetermined attribute in the contract data stored in the contract user database; 16. The information providing method according to claim 15, wherein the record for each user is generated as customer sending data. 17. The method according to claim 17, wherein the step of generating the customer transfer data includes the step of generating customer transfer data based on information indicating whether or not to display in the customer log stored in the customer log database and information indicating whether or not a contract has been made. The information providing method according to claim 15 or 16, wherein the information is generated. 18. The information providing method, comprising: updating a customer log stored in the customer log database based on the received contract information; updating contract information according to a predetermined category;
An information providing method, comprising: