Facility Information Creation System

Abstract

To provide a technique of reducing the possibility that facility information about a target facility is confused with facility information about another facility.SOLUTION: A facility information creation system includes: a facility keyword extraction unit that extracts a facility keyword relevant to a target facility on the basis of an article on the target facility; a category keyword extraction unit that extracts category keywords which are relevant to a target category of the target facility, and are the top N (N is an integer of 1 or greater) keywords having the highest relevance with the target category; a feature keyword extraction unit that extracts, as a feature keyword, the facility keyword which is the same as the category keyword and has a value indicating the relevance between the facility keyword and the target facility higher than a value indicating the relevance between the category keyword and the target category; and a facility information creation unit that creates facility information about the target facility by using the feature keyword.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a facility information creation system.

Background Art

[0002] Conventionally, a technique for extracting related keywords representing facility characteristics has been known by extracting frequently occurring keywords from posts on SNS related to facilities, reports in the media, and the like. Patent Document 1 describes extracting words (reference keywords) constituting a web page and calculating an evaluation value indicating the importance of the reference keywords on the web page.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] By the way, in facilities of the same category, similar keywords can become related keywords. Therefore, for example, even if introduction texts of each facility are created using such related keywords, there is a problem that it is difficult to understand the characteristics of each facility from such introduction texts. The present invention has been made in view of the above problems, and an object thereof is to provide a technique for reducing the possibility that the facility information of a target facility is confused with the facility information of other facilities.

Means for Solving the Problems

[0005] To achieve the above object, the facility information creation system includes a facility keyword extraction unit that extracts facility keywords related to a target facility based on an article about the target facility, a category keyword extraction unit that extracts category keywords that are keywords related to a target category to which the target facility belongs and are the top N (N is an integer of 1 or more) keywords having a high relevance to the target category, a feature keyword extraction unit that extracts, as feature keywords, facility keywords that are the same as the category keywords and for which the value indicating the relevance between the facility keyword and the target facility is higher than the value indicating the relevance between the category keyword and the target category, and a facility information creation unit that creates facility information of the target facility using the feature keywords.

[0006] That is, in the facility information creation system, the value indicating the relevance between each category keyword related to the target category to which the target facility belongs and the target category is compared with the value indicating the relevance between each keyword that is a facility keyword related to the target facility and is the same as the category keyword and the target facility. Then, a keyword for which the latter is higher in relevance than the former and is a value indicating high relevance is extracted as a feature keyword, and facility information of the target facility is created using the feature keyword. Therefore, it is possible to reduce the possibility that the facility information of the target facility is confused with the facility information of other facilities by the facility information created in this way.

[0007] To achieve the above object, the facility information creation system includes a facility keyword extraction unit that extracts facility keywords related to a target facility based on an article about the target facility, a category keyword extraction unit that extracts category keywords that are keywords related to a target category to which the target facility belongs and are the top N (N is an integer of 1 or more) keywords having a high relevance to the target category, a specific keyword extraction unit that extracts, as specific keywords, facility keywords that are not the same as the category keywords and for which the value indicating the relevance between the facility keyword and the target facility is higher than the value indicating the relevance between at least one category keyword and the target category, and a facility information creation unit that creates facility information of the target facility using the specific keywords.

[0008] That is, in the facility information creation system, among the category keywords related to the target category to which the target facility belongs, a facility keyword that is at least one value and indicates a higher relevance to the target facility than the value indicating the relevance to the target category and does not correspond to the category keyword is extracted as a specific keyword. Then, the facility information of the target facility is created using the specific keyword. Therefore, the facility information created in this way can reduce the possibility of confusing the facility information of the target facility with the facility information of other facilities.

Brief Description of the Drawings

[0009]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Embodiments for Carrying Out the Invention

[0010] Here, the embodiments of the present invention will be described in the following order. (1) Configuration of the facility information creation system: (2) Category keyword extraction process: (3) Characteristic keyword and specific keyword extraction process: (4) Other embodiments:

[0011] (1) Configuration of the facility information creation system: FIG. 1 is a block diagram showing the configuration of a facility information creation system 10 according to the present invention. In the present embodiment, the facility information creation system 10 is configured by a server computer and provides facility information to a user of a navigation system 50 as a client. There may be a plurality of clients of the facility information creation system 10, but FIG. 1 shows only one of them. The facility information created in the present embodiment is information indicating a facility (POI (Point Of Interest)) that can be set as a destination or the like of the navigation system 50.

[0012] The facility information creation system 10 includes a control unit 20 having a CPU, a RAM, a ROM, etc., a recording medium 30, and a communication unit 41, and the control unit 20 can execute programs stored in the recording medium 30 and the ROM. In the present embodiment, a facility information creation program 21 can be executed as this program. The communication unit 41 includes a circuit for communicating with the navigation system 50, and the control unit 20 can communicate with the navigation system 50 by the processing of the facility information creation program 21.

[0013] The navigation system 50 is a device having a navigation function to a destination. The navigation system 50 may be configured by an in-vehicle device or may be configured as a portable terminal such as a tablet or a smartphone. The navigation system 50 includes a communication unit 50a, a GNSS reception unit 50b, a control unit 50c, and a user I / F unit 50d. The control unit 50c includes a CPU, a ROM, a RAM, and a recording medium, and the CPU can execute various programs including a navigation program recorded in the ROM and the recording medium.

[0014] The communication unit 50a includes a communication circuit for wireless communication with other devices. The navigation system 50 can communicate with the facility information creation system 10 via the communication unit 50a. The user I / F unit 50d is an interface unit for the user to input instructions and provide various types of information to the user. The user I / F unit 50d includes a touch panel display (not shown), switches, speakers, etc. That is, the user I / F unit 50d includes an output unit for images and sounds and an input unit for user instructions.

[0015] The GNSS receiving unit 50b is a device that receives signals of the Global Navigation Satellite System. The GNSS receiving unit 50b receives radio waves from navigation satellites and outputs a signal for calculating the position of the navigation system 50 via an interface (not shown). By executing a navigation program, the control unit 50c acquires this signal and obtains the position of the navigation system 50.

[0016] The user of the navigation system 50 can specify search conditions including a search word and request a facility search from the facility information creation system 10. When the navigation system 50 obtains the facilities resulting from the facility search and the facility information of the facilities from the facility information creation system 10, it causes the user I / F unit 50d to display the searched facilities together with the facility information. Note that the trigger for presenting the facility information to the user may be other than a facility search. For example, when a facility displayed on a map is selected by the user, the facility information of the facility may be configured to be displayed.

[0017] The facility information creation system 10 has a function of creating facility information that can be presented to users as described above. In the recording medium 30 of the facility information creation system 10, an article DB 30a and a facility DB 30f are recorded. The facility DB 30f is a database that accumulates information about facilities. In the facility DB 30f, the name, location information, address, telephone number, and category to which the facility belongs are included in advance in association with the facility ID of each facility. The category indicates the category to which the facility belongs when categorized. The category may have a hierarchical structure. The facility DB 30f is used for facility search in the navigation system 50, recommendation of facilities, and the like.

[0018] The article DB 30a includes a plurality of article data (text data) indicating the contents of articles posted on blogs and SNS (Social Networking Services), and articles reported on TV, newspapers, magazines, and the like. The facility ID of the facility is associated with the article data that mentions a specific facility. The date and time when the article was published, such as the posting date or the reporting date, is associated with the article data. The facility information creation system 10 collects these article data in advance and accumulates them in the article DB 30a. The category keyword DB 30b, the facility keyword DB 30c, the feature keyword DB 30d, and the specific keyword DB 30e are generated when the control unit 20 executes the facility information creation program 21. Details will be described later.

[0019] The facility information creation program 21 includes a facility keyword extraction unit 21a, a category keyword extraction unit 21b, a feature keyword extraction unit 21c, a specific keyword extraction unit 21d, a facility information creation unit 21e, a search result acquisition unit 21f, and a search result display unit 21g. By the function of the search result acquisition unit 21f, the control unit 20 acquires the search results of the facility search based on the search word. That is, the control unit 20 acquires the search conditions of the facility search including the search word transmitted from the navigation system 50, and searches for facilities that match the search conditions with reference to the facility DB 30f. Note that the search process may be executed in another system that performs the search process, and the search result acquisition unit 21f may be configured to notify the system of the search conditions and acquire the search results from the system. Various known techniques may be adopted as the search algorithm. A rank is associated with the facilities obtained as the search results. The higher the relevance of the facility to the search conditions, the higher the rank.

[0020] By the function of the search result display unit 21g, the control unit 20 displays the search results. That is, the control unit 20 notifies the navigation system 50 of the facilities obtained as the search results, and causes the display of the user I / F unit 50d of the navigation system 50 to display them in accordance with the ranks of the search results. In the present embodiment, when the facilities obtained as the search results are displayed in accordance with the ranks, for each facility, facility information for introducing the facility together with the name of the facility or the like is displayed. In order to create facility information including the feature keyword or the specific keyword of the facility for each facility, the control unit 20 extracts the facility keyword for each facility and the category keyword for each category in advance, as will be described later. Then, based on the category keyword and the facility keyword, the feature keyword and the specific keyword are extracted for each facility.

[0021] Specifically, due to the function of the facility keyword extraction unit 21a, the control unit 20 extracts facility keywords related to the target facility based on the articles about the target facility. The target facility is an individual facility to be processed among a plurality of facilities, and here it is the facility targeted for extracting facility keywords. The control unit 20 refers to the article DB 30a and acquires the M most recent article data associated with the target facility. M may be any number of articles targeted for extracting facility keywords, and for example, about 100 may be assumed. The control unit 20 can identify the M most recent ones from the publication date and time of each article data and the current date and time. Then, the control unit 20 performs language analysis on the acquired M article data and extracts words. In this embodiment, the value indicating the relevance between a word and the target facility is represented by the value indicating the frequency of appearance of the word in all the acquired article data. For example, the word α is included in J (j = 1 to J) of the M article data K j (j = 1 to J), and the frequency of appearance of the word α in each article data K j is set as F j . Then, the value indicating the relevance between the word α and the target facility is represented by the sum from F1 to F J . Note that the frequency of appearance F j of the word α in the article data K j is, in this embodiment, represented by the value indicating the ratio of the frequency of the word α appearing in the article data K j to the total number of words in the article data K j . In this way, the control unit 20 extracts facility keywords for each facility recorded in the facility DB 30f and records them in the facility keyword DB 30c in association with the facility ID, together with the value indicating the relevance between each facility keyword and the target facility.

[0022] By the function of the category keyword extraction unit 21b, the control unit 20 extracts category keywords, which are the top N (N is an integer of 1 or more) keywords related to the target category to which the target facility belongs and have a high relevance to the target category. In the present embodiment, the control unit 20 acquires from the article DB 30a the article data including the word indicating the name of the target category and the article data associated with all the facilities belonging to the target category, and the most recent M pieces of article data. Although the same article may be acquired repeatedly by acquiring the article data including the word indicating the name of the target category and the article data associated with all the facilities belonging to the target category, when the same article data is acquired repeatedly, the control unit 20 eliminates the duplication and treats it as one piece of article data. The control unit 20 performs language analysis on the acquired M pieces of article data and extracts words. In the present embodiment, the value indicating the relevance between the word and the target category is represented by the value indicating the frequency of appearance of the word in the acquired M pieces of article data. The calculation method is the same as that in the case of the facility keyword. For each word, when calculating the value indicating the relevance to the target category, the control unit 20 sorts the words in descending order of the value and extracts the top N (N is set to 10 in the present embodiment) as the category keywords. The control unit 20 extracts category keywords in the same manner for all categories, and records the category keywords in the category keyword DB 30b in association with the categories, together with the values indicating the relevance between each category keyword and the target category.

[0023] By the function of the feature keyword extraction unit 21c, the control unit 20 extracts, as the feature keyword, a facility keyword that is the same as the category keyword and for which the value indicating the relevance between the facility keyword and the target facility is higher than the value indicating the relevance between the category keyword and the target category. To do so, the control unit 20 acquires, for each facility keyword that is the same as the N words extracted as the category keyword, the value indicating the relevance to the target facility. Then, for the same word, the value V1 indicating the relevance to the target facility and the value V2 indicating the relevance to the target category are compared, and if there is a word for which the value V1 is greater than the value V2, the word is acquired. In the present embodiment, further, when the value (difference) obtained by subtracting the value V2 from the value V1 is equal to or greater than the threshold value, the control unit 20 extracts the word as the feature keyword of the target facility.

[0024] FIG. 2 is a diagram showing an example of the relationship between the value indicating the relevance to the target category and the value indicating the relevance to an individual facility (target facility) belonging to the target category, for each category keyword of the target category. In this example, the target category is "shrine" and the target facility is "AAA Shrine". Assuming that the word for which the value (difference) obtained by subtracting the value V2 from the value V1 is equal to or greater than the threshold value is "goshuin", the control unit 20 extracts "goshuin" as the feature keyword. Similarly, the control unit 20 extracts the feature keyword for each facility recorded in the facility DB 30f and records it in the feature keyword DB 30d in association with the facility ID.

[0025] The feature keyword is a keyword that has a high relevance to the category to which the target facility belongs and is among the top N keywords in terms of relevance, and for which the relevance to the individual target facility is higher than the relevance to the entire target category. Therefore, the control unit 20 can extract, as the feature keyword, a keyword that has a high relevance to the entire category and also has a particularly high relevance to the individual target facility.

[0026] By the function of the specific keyword extraction unit 21d, the control unit 20 extracts, as specific keywords, facility keywords that are not the same as the category keywords and for which the value indicating the relevance between the facility keyword and the target facility is higher than the value indicating the relevance between at least one category keyword and the target category. In the present embodiment, the control unit 20 identifies the minimum value of the value (V2) indicating the relevance between the category keyword and the target category. Then, the control unit 20 extracts, as the specific keyword of the target facility, a facility keyword that is not the same as the category keyword, that is, the same as the keyword that did not rank in the top N among the keywords related to the target category, and for which the value (V1) indicating the relevance between the facility keyword and the target facility is greater than the above-mentioned minimum value.

[0027] Since the minimum value of the value V2 may vary depending on each category, in the present embodiment, no fixed threshold is provided, and the minimum value of the value V2 is used as the threshold (variable).

[0028] In the example of FIG. 2, the minimum value of the value V2 is 7, and the value V1 of "Nagoya", which is not a category keyword, is greater than the minimum value of the value V2. Therefore, the control unit 20 extracts "Nagoya" as the specific keyword of "AAA Shrine". Similarly, the control unit 20 extracts specific keywords for each facility recorded in the facility DB 30f and records them in the specific keyword DB 30e in association with the facility ID.

[0029] Specific keywords are facility keywords that have a high relevance to an individual target facility among the keywords that are not highly relevant to the category to which the target facility belongs to the extent of being in the top N. Therefore, the control unit 20 can extract, as specific keywords, keywords that do not overlap with keywords that are highly relevant to the entire category but are particularly relevant to an individual target facility.

[0030] By the function of the facility information creation unit 21e, the control unit 20 creates facility information of the target facility using at least one of the feature keyword, the specific keyword, and at least one from the top to the L-th position (L is an integer of 1 or more and N or less) of the category keywords. In the present embodiment, when guiding the facilities in the search results by the search result display unit 21g, the facility information created by the function of the facility information creation unit 21e is displayed. Therefore, when the control unit 20 acquires the facility ID of the facility obtained as a result of the search by the search result acquisition unit 21f, by the function of the facility information creation unit 21e, it refers to the facility DB 30f and specifies the category to which the facility with the facility ID belongs. Subsequently, the control unit 20 refers to the category keyword DB 30b and acquires the category keyword of the category. Subsequently, the control unit 20 refers to the feature keyword DB 30d and acquires the feature keyword of the facility with the facility ID, and refers to the specific keyword DB 30e and acquires the specific keyword of the facility with the facility ID. Further, the control unit 20 selects at least one keyword that does not overlap with the feature keyword among, for example, the top 5 category keywords. For example, the keyword with the highest rank that is not a feature keyword may be selected.

[0031] For example, when creating facility information for the "AAA Shrine" shown in the example of FIG. 2, facility information is created using the characteristic keyword of the "AAA Shrine", which is "omamori", the specific keyword of the "AAA Shrine", which is "Nagoya", and "shrine" as one of the category keywords. The facility information may be any information created to introduce to the user what kind of facility it is. For example, it may be simply composed by listing the characteristic keyword, the specific keyword, and one of the category keywords. Alternatively, from the article data associated with the facility ID of the facility, an article using the characteristic keyword or an article using the specific keyword may be extracted and used as the facility information. Also, an article created by applying the characteristic word or the specific word to the XXX and YYY parts of a pre-prepared article such as "In particular, it is a topic with XXX and YYY" may be used as the facility information. Or, an AI engine that generates an article within a specified number of characters including the keyword (characteristic keyword, category keyword other than the characteristic keyword, specific keyword) is used to generate an article using these keywords and used as the facility information. By creating the facility information using at least one (other than the characteristic keyword) from the top to the L-th place of the category keywords in order, it is possible to create facility information including general keywords highly relevant to the entire target category to which the target facility belongs.

[0032] As described above, according to the present embodiment, it is possible to create and present to the user facility information using characteristic keywords and specific keywords related to the target facility. Therefore, it is possible to present to the user characteristic or specific information about the target facility as compared with other facilities related to the category to which the target facility belongs. As a result, the possibility that the facility information of the target facility is confused with the facility information of other facilities can be reduced.

[0033] In addition, when the top K facilities (K is an integer of 2 or more) in the search results are only facilities in the same category according to the function of the search result display unit 21g, the control unit 20 also additionally displays facilities outside the same category that have the search word as a specific keyword. That is, the control unit 20 specifies the category to which each of the top K facilities among the plurality of facilities obtained by the facility search based on the search word by the search result acquisition unit 21f belongs. When all of the top K facilities belong to the same category, the control unit 20 refers to the specific keyword DB 30e, selects facilities outside the other category that have the search word as a specific word, and adds the facilities to be displayed in addition to the top K facilities in the search results.

[0034] For example, when all of the top K facilities obtained by the facility search using the search word "goshuincho" belong to the "shrine" category, the control unit 20 specifies facilities outside the "shrine" category that have "goshuincho" as a specific keyword. FIG. 3 is a diagram showing another example of the relationship between the value indicating the relevance to the target category and the value indicating the relevance to an individual facility (target facility) belonging to the target category for each category keyword of the target category. In this example, the target category is "bookstore" and the target facility is "BBB Bookstore". In the example of FIG. 3, the minimum value of the value V2 is 7, and the value V1 of "goshuincho", which is not a category keyword, is larger than the minimum value of the value V2. Therefore, in the example of FIG. 3, the specific keyword of "BBB Bookstore" is "goshuincho". Therefore, the control unit 20 also causes the navigation system 50 to display "BBB Bookstore" in addition to the top K facilities in the "shrine" category in the search results. For example, the control unit 20 creates a sentence such as "AA Shrine, CCC Shrine, and DDD Shrine are particularly popular. Outside of shrines, BBB Bookstore is also becoming a topic." as the search result for "goshuincho".

[0035] In this way, when all of the top K facilities in the search results belong to the same category, by also presenting facilities outside the other category that have the search word as a specific keyword, it is possible to make the user recognize the existence of facilities with unexpectedness.

[0036] (2) Category keyword extraction process: Next, the category keyword extraction process executed by the control unit 20 by the function of the category keyword extraction unit 21b will be described with reference to FIG. 4. The category keyword extraction process may be executed periodically, may be executed at any timing that is not periodic, or may be executed in response to a request for facility search. In this embodiment, it is assumed that the process is executed periodically, for example, once a week. When the category keyword extraction process is started, an article that mentions the target category is acquired (step S100). That is, the control unit 20 acquires the most recent M articles that include a word indicating the target category and are associated with facilities that belong to the target category, by referring to the article DB 30a. For example, if the target category is "shrine", the most recent M articles among those that include the word "shrine" and those that are associated with facilities that belong to the shrine category are acquired from the article DB 30a. The acquired articles may include articles about individual shrines, articles about multiple individual shrines, articles that do not mention individual shrines, and articles about facilities that belong to a category other than the shrine category and mention "shrine".

[0037] Next, the control unit 20 performs linguistic analysis on each of the acquired articles to extract words, and calculates the word relevance for each word (step S105). The control unit 20 divides the text data included in one article into words, and calculates the frequency of occurrence of each word in one article. In this embodiment, the article data K j Frequency of occurrence of word α in F j As, article data K j The frequency of the word α appearing in the article data K j A value indicating the proportion of each word in the article to the total number of words in the article is calculated. For example, a word that appears frequently in an article using the word "shrine" is considered to be highly related to the "shrine" category. The control unit 20 performs this process for each extracted word for all articles acquired in step S100.

[0038] Subsequently, the control unit 20 adds a value indicating the relevance of each extracted word in each article (step S110). For example, in step S105, for word α, the frequency of occurrence F of word α in each article data K j in which word α appears j is calculated, and all of them are added. If a word with a high frequency of occurrence in a certain article is also used in many other articles, this word can be regarded as highly relevant to the target category. Note that the control unit 20 normalizes the value calculated in this way by dividing it by the number of articles obtained in step S100.

[0039] Subsequently, the control unit 20 records the top N words in descending order of relevance as category keywords (step S115). That is, the control unit 20 records N category keywords and a value indicating the relevance of each category keyword to the target category in the category keyword DB30b in association with the target category. By performing the above category keyword extraction process for each category by the control unit 20, the category keywords for each category can be extracted.

[0040] (3) Feature keyword and specific keyword extraction process: The feature keyword and specific keyword extraction process executed by the control unit 20 will be described with reference to FIG. 5. The feature keyword and specific keyword extraction process may be executed periodically, at any non-periodic timing, or in response to a facility search request. In this embodiment, it is assumed that the process is executed periodically, for example, every week. When the feature keyword and specific keyword extraction process is started, the control unit 20 acquires the articles that mention the target facility by the function of the facility keyword extraction unit 21a (step S200). That is, the control unit 20 refers to the article DB 30a and acquires the most recent M articles among the article data recorded in association with the facility ID of the target facility. If the number of article data recorded in association with the facility ID of the target facility is less than M, all article data less than M are acquired in step S200.

[0041] Subsequently, the control unit 20 performs language analysis on each of the acquired articles by the function of the facility keyword extraction unit 21a to calculate the word relevance (step S205), and for each of the extracted words, adds the value indicating the word relevance calculated in each article (step S210). Steps S205 and S210 are the same processes as steps S105 and S110 of the category keyword extraction process in FIG. 4. The value calculated in step S210 is normalized by dividing by the number of articles acquired in step S200. As a result of the control unit 20 executing steps S205 and S210, a plurality of facility keywords for the target facility and values indicating the relevance for each of the facility keywords are calculated.

[0042] Subsequently, the control unit 20 performs the processes from step S215 to S240 for each word extracted in step S205. That is, one unprocessed word out of the plurality of words extracted in step S205 is sequentially selected, and the processes from step S215 to S240 are performed. The control unit 20 determines, by the function of the feature keyword extraction unit 21c, whether the target word has already been recorded as a category keyword of the category to which the target facility belongs (step S215). That is, the control unit 20 refers to the category keyword DB 30b, acquires the N category keywords of the category to which the target facility belongs, and determines whether the target word is identical to any of the N category keywords.

[0043] In step S215, if it is determined that the word has already been recorded, the control unit 20 calculates, by the function of the feature keyword extraction unit 21c, the difference between the value indicating the relevance of the target word and the value indicating the relevance of the category keyword identical to the target word (step S220). That is, the control unit 20 calculates value V1 - value V2. Subsequently, the control unit 20 determines, by the function of the feature keyword extraction unit 21c, whether the difference (V1 - V2) is equal to or greater than the threshold value (step S225). If it is determined in step S225 that the difference is equal to or greater than the threshold value, the control unit 20 records, by the function of the feature keyword extraction unit 21c, the target word as a feature keyword of the target facility (step S230). That is, the control unit 20 records the target word (feature keyword) in the feature keyword DB 30d in association with the target facility.

[0044] In step S215, when it is not determined that the target word has been recorded as a category keyword of the target facility's category, the control unit 20 determines, by the function of the specific keyword extraction unit 21d, whether the value indicating the relevance of the target word is greater than the value indicating the relevance of the lowest-level category keyword (step S235). If it is determined in step S235 that the former is greater than the latter, the control unit 20 records the target word as a specific keyword of the target facility by the function of the specific keyword extraction unit 21d (step S240). That is, the control unit 20 records the target word (specific keyword) in association with the target facility in the specific keyword DB 30e.

[0045] If it is not determined in step S225 that the difference is equal to or greater than the threshold value, or after step S230 is executed, or after step S240 is executed, or if it is not determined in step S235 that the former is greater than the latter, the control unit 20 determines whether processing has been completed for all the words for which relevance has been calculated (step S245). If it is not determined in step S245 that processing has been completed, the control unit 20 performs the processing from step S215 on the next target word.

[0046] By performing the above characteristic keyword and specific keyword extraction processes for each facility, it is possible to extract facility keywords, characteristic keywords, and specific keywords for each facility. When the facility search request from the navigation system 50 is acquired, the control unit 20 performs a facility search based on the search conditions including the search word by the search result acquisition unit 21f, and acquires the facility IDs of a plurality of facilities in the search result. The control unit 20 sets each of these facilities as a target facility, and acquires the characteristic keywords and specific keywords of the target facility by referring to the characteristic keyword DB 30d and the specific keyword DB 30e. Further, the control unit 20 acquires at least one (excluding the characteristic keyword) up to the top L bits of the category keywords of the category to which the target facility belongs by referring to the category keyword DB 30b. Then, the control unit 20 creates facility information using the acquired characteristic keyword, specific keyword, and at least one of the category keywords, and returns the facility information of the facility together with the facility IDs of the plurality of facilities obtained as a result of the facility search to the navigation system 50. As a result, in the navigation system 50, facility information is presented for each of the plurality of facilities obtained as a result of the facility search. The user can grasp the characteristic information and specific information for each facility based on this facility information.

[0047] (4) Other Embodiments: The above embodiments are examples for implementing the present invention, and various other embodiments can also be adopted. For example, the facility information creation system 10 may be configured by a plurality of devices (such as a server, a client, etc.). For example, the facility information creation unit 21e and the search result display unit 21g constituting the facility information creation system may exist in a client device. A part of the configuration of the above-described embodiments may be omitted, or the order of processing may be changed or omitted.

[0048] In the feature keyword extraction unit, if there is no word whose value (difference) obtained by subtracting value V2 from value V1 is greater than or equal to the threshold value, the word with the largest difference may be extracted as the feature keyword, or a predetermined number of words may be extracted as the feature keywords in descending order of the difference. Multiple feature keywords may be extracted, or only one feature keyword may be extracted.

[0049] The specific keyword is a keyword that does not correspond to the category keyword and is a facility keyword whose value indicating the relevance is greater than that of at least one category keyword. "At least one category keyword" is not limited to the category keyword with the minimum relevance value. Multiple specific keywords may be extracted, or only one specific keyword may be extracted. When there are multiple facility keywords that are not the same as the category keyword and whose value indicating the relevance between the facility keyword and the target facility is higher than the value indicating the relevance between at least one category keyword and the target category, a predetermined number of them may be extracted as specific keywords in descending order of the relevance value.

[0050] The value indicating the relevance to the target facility only needs to be able to indicate the level (strength, magnitude) of the relevance to the target facility, and various calculation methods may be adopted. Also, for the value indicating the relevance to the target category, it only needs to be able to indicate the level (strength, magnitude) of the relevance to the target category, and various methods may be adopted. For calculating the value indicating the relevance, indicators other than frequency may be used. For example, the value indicating the relevance of the word specified in advance as a keyword by the article author may be set as the value indicating high relevance. Known techniques related to natural language processing and keyword extraction processing may be adopted. The relationship between different multiple words in one article (for example, a strong tendency to appear together in the same article, etc.) may be used as an indicator for calculating the value indicating the relevance. Also, the distributed representation of words may be defined, and the relevance may be defined based on the distance in the vector space, etc.

[0051] The facility information may be configured to be created using the characteristic keywords of the target facility without using specific keywords. Alternatively, the facility information may be configured to be created using, without using the specific keywords of the target facility, at least one (other than the characteristic keyword) of the characteristic keyword and the category keywords from the top in order up to the L-th position (L is an integer of 1 or more and N or less) of the category keywords to which the target facility belongs. Further, the facility information may be configured to be created using specific keywords without using the characteristic keywords of the target facility. Further, the facility information may be configured to be created using, without using the characteristic keywords of the target facility, at least one category keyword up to the L-th position in order from the top of the specific keyword of the target facility and the category keywords to which the target facility belongs.

[0052] In the above embodiment, when the value obtained by subtracting the value V2 from the value V1 (the difference) is equal to or greater than the threshold value, the control unit 20 is configured to extract the word as a characteristic keyword of the target facility. However, without setting a threshold value, it may be configured to obtain a predetermined number of facility keywords in descending order of values as characteristic keywords when the difference (V1 - V2) is positive.

[0053] When the search word is a category keyword of a specific category, the search display unit may be configured to limit the number of displayed facilities of the specific category in the search results and display, as search results, facilities of categories other than the specific category for which the search word is a specific keyword together with the facilities of the specific category with the limited number of displays. When the search word is a category keyword of a specific category, it is expected that the search results will be biased towards the facilities of the specific category. Therefore, by limiting the number of displays of the search results of the specific category and, instead, also displaying facilities of categories other than the specific category for which the search word is a specific keyword in the search results, it is possible to make the user recognize the existence of facilities that are unexpected.

[0054] Furthermore, the method of the present invention is also applicable as a program or a method. Also, the above-described system, program, and method may be realized as a single device, or may be realized by using components shared with each part provided in a vehicle, and include various aspects. Also, it can be appropriately changed, such as partly being software and partly being hardware. Furthermore, the invention is also established as a recording medium for a program that controls the system. Of course, the recording medium for the program may be a magnetic recording medium, a semiconductor memory, or any recording medium developed in the future, and can be considered in exactly the same way.

Explanation of Reference Numerals

[0055] 10... Facility information creation system, 20... Control unit, 21... Facility information creation program, 21a... Facility keyword extraction unit, 21b... Category keyword extraction unit, 21c... Feature keyword extraction unit, 21d... Specific keyword extraction unit, 21e... Facility information creation unit, 21f... Search result acquisition unit, 21g... Search result display unit, 30... Recording medium, 30a... Article DB, 30b... Category keyword DB, 30c... Facility keyword DB, 30d... Feature keyword DB, 30e... Specific keyword DB, 30f... Facility DB, 41... Communication unit, 50... Navigation system, 50a... Communication unit, 50b... GNSS receiver, 50c... Control unit, 50d... User I / F unit

Claims

1. A facility keyword extraction unit that extracts facility keywords related to the target facility based on an article about the target facility, A category keyword extraction unit that extracts category keywords, which are keywords related to the target category to which the target facility belongs and are the top N (N is an integer of 1 or more) keywords with high relevance to the target category, A feature keyword extraction unit that extracts the facility keyword that is the same as the category keyword and for which the value indicating the relevance between the facility keyword and the target facility is higher than the value indicating the relevance between the category keyword and the target category, and uses the facility keyword as a feature keyword, A facility information creation unit that creates facility information of the target facility using the feature keyword, A facility information creation system comprising the above.

2. A facility keyword extraction unit that extracts facility keywords related to the target facility based on an article about the target facility, A category keyword extraction unit that extracts category keywords, which are keywords related to the target category to which the target facility belongs and are the top N (N is an integer of 1 or more) keywords with high relevance to the target category, A specific keyword extraction unit that extracts the facility keyword that is not the same as the category keyword and for which the value indicating the relevance between the facility keyword and the target facility is higher than the value indicating the relevance between at least one of the category keywords and the target category, and uses the facility keyword as a specific keyword, A facility information creation unit that creates facility information of the target facility using the specific keyword, A facility information creation system comprising the above.

3. Comprising a specific keyword extraction unit that extracts the facility keyword that is not the same as the category keyword and for which the value indicating the relevance between the facility keyword and the target facility is higher than the value indicating the relevance between at least one of the category keywords and the target category, The facility information is created using the feature keyword and the specific keyword, The facility information creation system according to Claim 1.

4. The facility information is created using at least one of the top L (L is an integer of 1 or more and N or less) from the top of the category keywords, The facility information creation system according to Claim 3.

5. A search result acquisition unit that acquires search results obtained by searching for facilities based on a search keyword, When the top K (K is an integer of 2 or more) of the search results are only facilities in the same category, a search result display unit that additionally displays facilities that are facilities other than the same category and for which the search keyword is the specific keyword, The facility information creation system according to any one of claims 2 to 4, comprising:

Images