JP2014235597A - Specific point name determination apparatus, specific point name determination method, and specific point name determination program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve a recall ratio of extraction of a specific point name used as one meaning a specific point in a text.SOLUTION: A specific point name determination apparatus 1A comprises: a specific point information storage unit 13 storing therein a specific point name and specific point information; a feature word storage unit 15A storing therein a feature word characterizing that the specific point name means a specific point in a text per specific point; a text extraction unit 14 extracting a determination target text; a determination unit 18A determining that the specific point name means the specific point in the determination target text if a first flag is on to correspond to the specific point name included in the determination target text and the feature word stored in the feature word storage unit 15A so as to be associated with the specific point name is included in the determination target text; and a feature word generation unit 20 storing a feature word generated based on the determination target text determined by the determination unit 18A in the feature word storage unit 15A.

Description

  The present invention relates to a specific point name determination apparatus, a specific point name determination method, and a specific point name determination program for determining a specific point name in a text.

  Conventionally, as a technique for extracting notation (specific point name) indicating a specific point such as a place name / facility name from a text, a technique shown in Non-Patent Document 1 below is known. The basic method of such a technique is to prepare a place name dictionary in which a specific point name is stored in advance, and to detect a specific point name in the text by comparing the word in the text with the place name dictionary. Is.

  By the way, in the specific point name, there is a specific point name that is used as a place name as well as a personal name, such as “Oyama”. In addition, there is a specific point name such as “Kiyomizu Temple” where facilities having the same name are scattered in a plurality of different areas (Kyoto City, Osaka City, etc.). Therefore, it may not be possible to appropriately determine whether or not the specific point name in the text is used to mean a specific point such as a place name or facility name only by dictionary-based collation as described above. In addition, it may not be possible to appropriately determine which specific point the specific point name in the text indicates.

  In the following Non-Patent Document 1, in order to eliminate such ambiguity of the specific point name and appropriately identify and extract the specific point name in the text, the specific point name to be extracted should be used as a person name. Personal name category indicating that there are multiple names, same-name categories indicating that there are multiple place names / facility names, etc., and general term categories indicating that they may be used in general terms such as product names and nicknames Classification into And for specific point names classified into the same name category, based on the combination of the specific point name appearing in the text and the surrounding place name corresponding to the specific point name, It is determined whether it is what is shown. In addition, for a specific point name classified into the personal name category or general language category, if the name of the surrounding place corresponding to the specific point indicated by the specific point name is not included in the text, the specific point name is the place name / facility name, etc. It is determined that it is not used in the sense of.

Einat Amitay, Nadav Har’El, Ron Sivan, Aya Soffer, “Web-a-Where: Geotagging Web Content”, ACMSIGIR 2004, Pages 273-280

  By the way, in the above-mentioned processing method, since the conditions for determining that the specific point name in the text means the specific point are severe, there is a problem that the precision (accuracy) is high but the reproduction rate is low. In particular, when a specific point name determination process is performed on a tweet (short text) posted on a microblog service such as Twitter (registered trademark) using the above-described accuracy-oriented processing method, text There is a high possibility that the name of a specific point used to mean a specific point is not extracted at all.

  Therefore, in view of the above problems, the present invention provides a specific point name determination device, a specific point name determination method, and a specific point name determination method that can increase the reproduction rate of specific point names used as meaning specific points in text. The purpose is to provide a specific point name determination program.

  The specific point name determination apparatus according to the present invention relates to a specific point name indicating a predetermined specific point name and a first flag indicating that the specific point name has a meaning other than the specific point. Specific point information storage means for storing point information, characteristic word storage means for storing a characteristic word for each specific point, which is a word characterizing that the specific point name means a specific point in the text, and text for acquiring text Corresponding to the acquisition means, the text extraction means for extracting the determination target text including the specific spot name associated with the specific spot information from the text acquired by the text acquisition means, and the specific spot name included in the determination target text The first flag indicates that the specific point name has a meaning other than the specific point, and is associated with the specific point name. If the feature word stored in the memory means is included in the determination target text, the determination unit determines that the specific point name means the specific point in the determination target text, the determination target text, and the determination target A feature word generating unit that generates a feature word based on determination result data in which a determination result by the determination unit for the text is associated, and stores the generated feature word in the feature word storage unit.

  The specific point name determination method according to the present invention relates to a specific point name indicating a predetermined specific point name and a first flag indicating that the specific point name has a meaning other than the specific point. Specific point name storage means for storing point information, and specific word name storage means for storing, for each specific point, a characteristic word that is a word characterizing that the specific point name means a specific point in the text A specific point name determination method executed by a determination device, wherein a text acquisition step for acquiring text and a determination target text including a specific point name associated with specific point information are acquired in the text acquisition step The text extraction step to extract from the first flag corresponding to the specific point name included in the determination target text If the point name has a meaning other than the specific point and the characteristic word stored in the characteristic word storage means in association with the specific point name is included in the determination target text, the determination target text A feature word is generated and generated based on determination result data that associates a determination step that determines that the specific point name means a specific point, and a determination target text and a determination result in the determination step for the determination target text. And a feature word generation step of storing the feature word in the feature word storage means.

  The specific point name determination program according to the present invention causes the computer to include a specific point name indicating a name of a predetermined specific point and a first flag indicating that the specific point name has a meaning other than the specific point. Specific point information storage means for storing the related specific point information, characteristic word storage means for storing a characteristic word for each specific point, which is a word characterizing that the specific point name means a specific point in the text, and text A text acquisition means for acquiring, a text extraction means for extracting the determination target text including the specific spot name associated with the specific spot information from the text acquired by the text acquisition means, and a specific spot name included in the determination target text The first flag corresponding to indicates that the specific point name has a meaning other than the specific point. A determination unit that determines that the specific point name means a specific point in the determination target text if the characteristic word stored in the characteristic word storage unit in association with the name is included in the determination target text; A feature word is generated based on determination result data in which the target text and a determination result by the determination unit for the determination target text are associated with each other, and the generated feature word is stored in the feature word storage unit.

  In this invention, the text extraction means refers to the specific spot information storage means and extracts the determination target text including the specific spot name from the text acquired by the text acquisition means. Subsequently, when the specific point name included in the determination target text has a meaning other than the specific point by the first flag, the determination unit stores the characteristic point storage unit in association with the specific point name. Whether or not the specific point name in the determination target text means a specific point is determined based on whether or not the feature word being included is included in the determination target text. Subsequently, the feature word generation unit generates a feature word based on the determination result data in which the determination target text is associated with the determination result by the determination unit, and stores the generated feature word in the feature word storage unit. By repeating this series of processing, the feature words stored in the feature word storage unit increase, and therefore, in the determination by the determination unit, the specific point name meaning the specific point in the determination target text is appropriately extracted ( The possibility that it can be determined to mean a specific point) can be increased. That is, it is possible to increase the recall of the extraction of the specific point name used as meaning the specific point in the text.

  The specific point name determination apparatus further includes a determination model storage unit that stores a determination model that inputs a determination target text and determines whether the specific point name means a specific point in the determination target text. The word storage means stores an incorrect answer feature word, which is a word characterizing that the specific point name does not mean a specific point in the text, together with the characteristic word for each specific point name, and the determination unit determines whether to determine based on the determination model When it is determined that the specific point name included in the text does not mean the specific point in the determination target text, the characteristic word stored in the characteristic word storage unit in association with the specific point name is included in the determination target text If it is determined that the specific point name in the determination target text means a specific point, the determination target text is determined based on the determination model. In the case where it is determined that the specific point name included in the text means the specific point in the determination target text, the incorrect characteristic word stored in the characteristic word storage means in association with the specific point name is included in the determination target text If not, it is determined that the specific point name in the determination target text means a specific point, and the feature word generation unit generates a feature word or an incorrect feature word based on the determination result data, and the generated feature A word or an incorrect answer feature word may be stored in the feature word storage means.

  According to this specific point name determination apparatus, the determination using the determination model and the determination using the feature word or the incorrect feature word are performed together to specify the specific point used in the text. It is possible to increase the recall rate of the spot name extraction and increase the accuracy (matching rate). Specifically, even if it is determined that the specific point name does not mean the specific point in the determination target text by the determination by the determination model, the feature word characterizing that the specific point name means the specific point is If it is included, it is determined that the specific point name in the determination target text means a specific point. By such determination, the reproduction rate of the extraction of the specific point name is increased. On the other hand, if it is determined by the determination model that the specific point name means a specific point in the determination target text, an incorrect feature word that characterizes that the specific point name does not mean the specific point is not included Therefore, it is determined that the specific point name in the determination target text means a specific point. Such determination can improve the accuracy of extraction of the specific spot name.

  In the specific point name determination apparatus, the specific point information is associated with a second flag indicating that there are a plurality of the same name specific points having the same name as the specific point name associated with the specific point information. And a famousness storage unit that stores a pre-calculated famousness for each of the same-name specific points, and the determination unit includes a second flag corresponding to the specific point name included in the determination target text. In the case where it is indicated that there are a plurality of the same name specific points, it may be determined whether or not to execute the same name determination process based on the famousness of the same name specific points stored in the famousness storage means. . Here, based on a character string other than the specific point name included in the determination target text, the same name determination process indicates which specific point among the same name specific points in the determination target text. This is a process for determining.

  Conventionally, for text including a specific point name having a plurality of specific points with the same name, it is determined by the same name determination process shown below whether the specific point name indicates a specific point or whether or not a specific point is indicated. It is common. As an example, a specific point name “Kiyomizu-dera” existing in a plurality of areas such as Kyoto and Osaka will be described. In the same name determination process, if the surrounding place name (for example, “Kyoto”) corresponding to the specific point name is included in the text including the specific point name “Kiyomizu Temple”, the specific point name is changed to “Kiyomizu Temple in Kyoto”. ". On the other hand, if no surrounding place name corresponding to the specific location name is included in the text, the specific location name is not determined to indicate the specific location in the first place. Since the same name determination process determines that the specific spot name indicates a specific specific spot when the surrounding place name is included in the same text, the accuracy is high but the recall is low.

  In this specific point name determination device, when there are a plurality of specific points (same name specific points) having the same name as the specific point name included in the determination target text, the determination means determines the same name based on the famousness of the same name specific point Whether to execute the process can be switched. That is, for example, when the specific point indicated by the specific point name in the determination target text can be uniquely determined based on the famousness, the determination unit omits the same name determination process, and the specific point name is set to the famousity. It can be determined that the specific point determined based on this is shown. As a result, even when the surrounding text corresponding to the specific spot name is not included in the determination target text, the specific spot name can be appropriately extracted based on the famousness of the specific spot. The rate is improved.

  In the specific point name determination apparatus, the determination unit includes the determination unit in the determination target text without performing the same name determination process when the famous point of the specific point having the highest famousness among the same name specific points is equal to or greater than a predetermined threshold. It is determined that the specific point name included in indicates the specific point having the maximum famousness, and the famous point of the specific point having the maximum famousness among the specific names with the same name is less than a predetermined threshold, the same name determination process May be executed.

  According to this specific point name determination apparatus, when the famous point of the specific point having the highest famousness among the same name specific points is equal to or greater than a predetermined threshold, the same name determination process is omitted and the specific point name is appropriately extracted. Since this can be done, the recall can be improved.

  In the specific point name determination device, the determination means has a degree of famousness that is less than or equal to a predetermined value when the famous point of the specific point having the highest famousness among the same name specific points is less than a predetermined threshold. The same name determination process may be executed by excluding the same name specific point.

  According to this specific point name determination apparatus, when the famous point of the specific point having the highest famousness among the same name specific points is less than a predetermined threshold, the specific point with the famous name equal to or lower than a predetermined value is excluded in advance. The same name determination process can be performed. Thereby, extraction of a specific point name can be performed efficiently and accurately.

  In the specific point name determination device, the number of surrounding place name texts including both the specific point and the surrounding place name corresponding to the specific point, the feature word including both the specific point and the characteristic word corresponding to the specific point Based on at least one of the number of texts and the number of check-ins indicating the number of times the user of the mobile communication terminal stayed at the specific point, the famous point of the specific point is calculated, and the calculated famous degree is stored in the famous degree storage means. You may further provide the famousness calculation means to memorize | store.

  According to this specific point name determination apparatus, the famousness calculation means can appropriately calculate the famousity of a specific point based on the number of surrounding place name texts, the number of feature word texts, the number of check-ins, and the like. . And the determination part determines the specific spot name based on the famousity calculated in this way, so that the accuracy of extracting the specific spot name can be improved.

  ADVANTAGE OF THE INVENTION According to this invention, the reproduction rate of extraction of the specific point name currently used as what means a specific point in a text can be raised.

It is a block diagram which shows the function structure of the specific point name determination apparatus which concerns on 1st Embodiment of this invention. It is a block diagram which shows the hardware constitutions of a specific point name determination apparatus. It is a figure which shows an example of specific point information. It is a figure used in order to explain the feature word memorized by the feature word storage part. It is a figure which shows an example of the famous degree of a specific point. It is a flowchart which shows operation | movement of the specific point name determination apparatus which concerns on 1st Embodiment. FIG. 7 is a flowchart showing personal name / general word flag processing shown in FIG. 6. It is a flowchart of the same name flag process shown in FIG. It is a figure which shows the function structure of the specific point name determination apparatus which concerns on 2nd Embodiment. It is a flowchart which shows the person name / general word flag process of the specific point name determination apparatus which concerns on 2nd Embodiment. It is a block diagram which shows the function structure of the specific point name determination program which concerns on one Embodiment of this invention.

  Embodiments according to the present invention will be described below with reference to the drawings. Where possible, the same parts are denoted by the same reference numerals, and redundant description is omitted.

[First Embodiment]
FIG. 1 is a block diagram showing a functional configuration of a specific point name determination apparatus 1A according to the first embodiment of the present invention. In the specific point name determination apparatus 1A according to the present embodiment, the specific point name means the specific point in the text including the specific point name (specific point name) (the specific point name is specified in the context of the text). It is a device that determines whether or not a point is used. Here, as the specific point, for example, a POI (Point of interest) indicating a specific place such as a place name or a facility name can be cited. Specifically, the specific point indicates a specific place such as a facility such as a temple or shrine (for example, “Kiyomizu Temple”), a mountain (for example, “Mt. Hakodate”), or a lake (for example, “Onuma”). As shown in FIG. 1, the specific point name determination device 1A includes a text acquisition unit 11, a text storage unit 12, a specific point information storage unit 13, a text extraction unit 14, a feature word storage unit 15A, a famousness calculation unit 16, and a famous name. A degree storage unit 17, a determination unit 18A, a determination result storage unit 19, and a feature word generation unit 20A are provided.

  FIG. 2 is a block diagram illustrating a hardware configuration of the specific point name determination apparatus 1A. As shown in FIG. 2, the specific point name determination apparatus 1A shown in FIG. 1 physically includes one or a plurality of CPUs 101, a main memory RAM 102 and ROM 103, and input devices such as a keyboard and a mouse. The computer system includes an input device 104, an output device 105 such as a display, a communication module 106 that is a data transmission / reception device such as a network card, and an auxiliary storage device 107 such as a hard disk drive and a semiconductor memory.

  Each function of the specific point name determination apparatus 1A in FIG. 1 is such that predetermined computer software is loaded on hardware such as the CPU 101 and RAM 102 shown in FIG. 105 and the communication module 106 are operated, and data is read and written in the RAM 102 and the auxiliary storage device 107. Hereinafter, each functional block will be described based on the functional blocks shown in FIG.

  The text input unit 2 is, for example, a post text posted to a microblog service from a server (not shown) that manages a microblog service such as Twitter (registered trademark) (“tweet” in the case of Twitter). ”) Is input (acquired) and one or more input post texts are stored. In the present embodiment, the text input unit 2 is realized as a device different from the specific point name determination device 1A. However, the text input unit 2 may be included in the specific point name determination device 1A. .

  The text acquisition unit 11 is a text acquisition unit that acquires post text from the text input unit 2. The text acquisition unit 11 acquires the posted text from the text input unit 2 periodically or in real time using, for example, a streaming API. The posted text acquired by the text acquisition unit 11 is stored in the text storage unit 12 such as a database. Here, the connection form between the text input unit 2 and the text acquisition unit 11 may be anything. For example, the text input unit 2 and the text acquisition unit 11 may be connected by a wireless method or may be connected by a wired method. The text input unit 2 and the text acquisition unit 11 may be connected (logic connection) via an arbitrary communication network such as the Internet, or may be connected by directly connecting devices. Good.

  The specific point information storage unit 13 is a specific point information storage unit such as a database that stores specific point information about a specific point indicating a predetermined specific place (POI or the like) for each specific point. It is assumed that the specific point information stored in the specific point information storage unit 13 is registered in advance by, for example, an administrator of the specific point name determination apparatus 1A. FIG. 3 is a diagram illustrating an example of specific point information. As shown in FIG. 3, the specific point information includes identification information (id) for uniquely specifying the specific point, a specific point name indicating the name of the specific point, the address of the specific point, and the position (latitude / longitude) of the specific point. An area code indicating an area (for example, “prefecture”, etc.) to which the specific point belongs, various flags (same name flag, personal name flag, general language flag), and the like are associated with each specific point.

  Here, the same name flag is a flag (second flag) indicating whether or not there are a plurality of specific points (same name specific points) having the same name as the specific point name. The personal name flag is a flag indicating whether or not the specific point name is the same name as the personal name. The general language flag is a flag indicating whether or not the specific point name is the same name as the general language. That is, the personal name flag and the general language flag are flags (first flag) indicating whether or not the specific point name has a meaning (person name and general language) other than the specific point. These flags are set to “1” when applicable and “0” when not applicable. In the following description, the flag is set to indicate that the flag is set to “1”.

  Various flags will be described in detail with reference to FIG. For example, since a specific point whose specific point name is “Kiyomizu-dera” exists in a plurality of lands such as Kyoto and Osaka, the same name flag is set. For the specific point with the specific point name “Onuma”, the same name flag is set because the specific point with the same name exists in a plurality of lands. In addition, the specific point name “Onuma” is sometimes used as a person's name (Mr. Onuma) and may be used as a general word (large swamp), so both the personal name flag and the general word flag are set.

  The text extraction unit 14 is a text extraction unit that extracts a posted text (determination target text) including a specific point name associated with the specific point information stored in the specific point information storage unit 13. For example, the text extraction unit 14 reads the specific point information stored in the specific point information storage unit 13 and loads it on the RAM 102. Subsequently, the text extraction unit 14 searches the text including the specific point name associated with the specific point information loaded on the RAM 102 from the plurality of posted texts stored in the text storage unit 12. Get the text to be judged. For example, when specific point information related to “Kiyomizu-dera” is loaded on the RAM 102, the text extraction unit 14 reads “Kiyomizu-dera” from the text storage unit 12 like “I am now sightseeing in Kiyomizu-dera”. The post text that contains the character string is extracted as the text to be judged. In the present embodiment, the text extraction unit 14 extracts one post text as a determination target text from among a plurality of post texts. However, the text extraction unit 14 includes, for example, one long text. A part including the specific point name (for example, one sentence or paragraph including the specific point name) may be extracted as the determination target text.

  The feature word storage unit 15A is a feature word storage unit that stores, for each specific point, a characteristic word that is a word characterizing that the specific point name means a specific point in the text. In the feature word storage unit 15A, a feature word (hereinafter referred to as “static feature word”) extracted in advance by an administrator or the like based on one or more sentences in which the specific point name is used as meaning the specific point in the text. Is also stored as a feature word of the specific point. Examples of such sentences include sentences described on Wikipedia and a homepage regarding a specific point. Static feature words are extracted by using, for example, tfidf. For example, as a feature word for a specific location called “Amanohashidate”, the name of the specific location “Amanohashidate” such as “beautiful” or “Miyazu” is used in the text to mean a specific location (that is, “Amanohashidate” as a place name) The words that characterize the fact that However, the feature words to be extracted may differ depending on which sentence is used as a source of feature word extraction by tfidf or the like. In addition to the static feature words, the feature word storage unit 15 also stores feature words generated by a feature word generation unit 20A described later.

  The famousness calculation unit 16 is a famousity calculation unit that calculates the famousness for each specific point of the same name and stores the calculated famousness in the famousness storage unit 17 such as a database. The famousness storage unit 17 is a famousity storage unit that stores the famousity calculated by the famousness calculation unit 16 for each of the specific points having the same name. The famousness storage unit 17 has a table area for storing the famousness for each specific point with the same name. Such data of the same name specific point (a list of the same name specific points having the same specific point name and the name of the city where each same name specific point exists) is stored in the public information such as Wikipedia and the specific point information storage unit 13, for example. By referring to the stored specific point information or the like, it is stored in the famousness storage unit 17 by an administrator or the like in advance.

  The celebrity calculation unit 16 includes the number of surrounding place name texts including both the same name specifying point and the surrounding place name corresponding to the same name specifying point, and both the same name specifying point and the feature word corresponding to the same name specifying point. Based on at least one of the number of feature word texts to be checked and the number of check-ins indicating the number of times the user of the mobile communication terminal has stayed at the same name specific point, the famous degree of the same name specific point can be calculated.

  Specifically, the famousness calculation unit 16 counts the number of surrounding place name texts and feature word texts by, for example, counting the texts acquired by the text acquisition unit 11 (“tweets” in the case of Twitter). Numbers can be aggregated. In addition, the famousity calculation unit 16 determines the number of check-ins for each specific point held by a social networking service (SNS) based on location information such as Foursquare, via an API published by the SNS. Can be acquired. Here, for example, it is assumed that table information in which peripheral place names are associated with each specific point with the same name is prepared in advance on a database (not shown) that can be referred to from the famousness calculation unit 16 by an administrator or the like. By referring to such a database, the famousness calculation unit 16 can total the number of surrounding place name texts.

  FIG. 5 is a diagram illustrating an example of the famousness for each specific point with the same name regarding the specific point whose specific point name is “Kiyomizu Temple”. Here, as an example, the famous number is the number of texts (TW number) obtained by adding the number of surrounding place name texts and the number of feature word texts. Here, “Kiyomizu Temple_city name” is a specific point indicating “Kiyomizu Temple existing in the city indicated by the city name”.

  The determination unit 18A is a determination unit that executes a determination process according to the state of a flag (same name flag, person name flag, general word flag) corresponding to a specific spot name included in the determination target text extracted by the text extraction unit 14. is there. Specifically, if the same name flag (second flag) is set, the determination unit 18A determines which same name specific point the specific point name in the determination target text indicates (same name flag process). Execute. On the other hand, if the personal name flag or the general word flag (first flag) is set, the determination unit 18A indicates whether the specific point name in the determination target text means a specific point, or whether it means a personal name or a general word Is executed (person name / general word flag processing).

  First, the same name flag process will be described. The determination unit 18 </ b> A determines whether or not to execute the same name determination process based on the famousness of the same name specific point stored in the famousness storage unit 17. The determination unit 18A is, for example, a specific point (hereinafter referred to as a determination target specific point) having the highest famousity among specific point names having the same name as a specific point name included in the determination target text (hereinafter referred to as “determination target specific point name”). Based on the famousity of the specific name with the same name as the specific name with the highest name, it is also known as “famous specific point”). It is determined whether or not it indicates a specific point. For example, the determination unit 18A determines whether the famousness of the famous specific point is equal to or greater than a predetermined threshold. If the famousness of the famous specific point is equal to or greater than the predetermined threshold, the determination target identification is performed in the determination target text. It is determined that the point name indicates a famous specific point (the determination target specific point name is used to mean a famous specific point). Moreover, the determination part 18A will perform the following same name determination processes, if the famousness degree of a famous specific point is less than a predetermined threshold value. Here, the predetermined threshold is set to an arbitrary value by an administrator or the like. The predetermined threshold is held as a setting file that can be read by the determination unit 18A, for example.

  With the same name determination process, based on a character string other than the specific point name included in the determination target text, it is determined which specific point among the same name specific points in the determination target text. It is processing. More specifically, in the determination target text, the same name determination process includes, for example, a determination target specific spot name and a surrounding place name (or a feature word, the nearest station name, etc.) corresponding to the specific point indicated by the determination target specific spot name. If included, it is a process for determining that the determination target specific point name means a specific point corresponding to the name of a surrounding place (or a feature word, the nearest station name, etc.). In the same-name determination process, for example, if “Kyoto”, which is one of the surrounding place names corresponding to the determination target specific point name, is included in the determination target text including “Kiyomizu Temple” which is the determination target specific point name. The determination target specific point name is determined to indicate “Kiyomizu Temple in Kyoto”. On the other hand, the determination target specific point name is not determined to indicate the specific point unless the determination target text includes any one of the surrounding place names corresponding to the specific point indicated by the determination target specific point name.

  For example, the determination unit 18A determines whether or not the name of the determination target specific point is famous by determining whether the ratio of the famousness of the famous specific point (the ratio of the famousness of all the same name specific points) is equal to or greater than a predetermined threshold. It can be determined whether or not it indicates a specific point.

  For example, let us consider a case where the determination target specific point name is “Kiyomizu Temple” and the predetermined threshold is 70%. In this case, in the example of FIG. 5A, the number of TW at a specific point indicating Kiyomizu-dera in Kyoto City is the largest and occupies approximately 88% of the total, so the determination unit 18A determines that the determination target specific point name is Kiyomizu-dera in Kyoto City It is determined that In the example of FIG. 5 (b), the TW number of the specific point indicating Kiyomizu-dera in Kyoto City is the largest and accounts for about 76% of the whole. Therefore, the determination unit 18A determines the name of the specific point to be determined for Kiyomizu-dera in Kyoto City. It is determined to be shown.

  Further, in the determination unit 18A, for example, the difference between the famousness of the specific point with the first (maximum) famousness and the famousness of the second specific point occupies the sum of the famousities of all the specific points with the same name. It may be determined whether or not the determination target specific point name indicates a famous specific point by determining whether or not the ratio is equal to or greater than a predetermined threshold.

  For example, let us consider a case where the determination target specific point name is “Kiyomizu Temple” and the predetermined threshold is 70%. In this case, in the example of FIG. 5 (a), Kiyomizu-dera Temple in Kyoto City with the first TW number and Kiyomizu-dera Temple in Isumi City with the second TW number with respect to the total famousity (200) of all the specific points with the same name. The ratio occupied by the difference in the number of TWs (176) is approximately 88%. Therefore, the determination unit 18A determines that the determination target specific point name indicates Kiyomizu Temple in Kyoto.

  On the other hand, in the example of FIG. 5 (b), Kiyomizu-dera in Kyoto City with the highest TW number and Kiyomizu-dera Temple in Isumi City with the second highest TW number with respect to the total famousity (241) of all the specific points with the same name. The ratio occupied by the difference in TW number (135) is approximately 56%. Therefore, the determination unit 18A performs the same name determination process without determining that the determination target specific point name indicates Kiyomizu-dera in Kyoto. Here, the determination unit 18 </ b> A may execute the same name determination process by excluding the same name specific points whose famousities are equal to or less than a predetermined value among the same name specific points. For example, the determination unit 18A excludes a specific point having the same name that is equal to or less than a predetermined value (here, “10” as an example), and the surrounding place name or feature word corresponding to Kiyomizu-dera in Isumi City with the second highest TW number is the text to be determined Only the determination of whether or not it is included is executed. Here, the determination unit 18 </ b> A determines that the determination target specific point name indicates Kiyomizu Temple in Isumi City when it is determined that the surrounding place name or feature word corresponding to Kiyomizu Temple in Isumi City is included in the determination target text. On the other hand, if the determination unit 18A determines that the surrounding place name or feature word corresponding to Kiyomizu-dera in Isumi city is not included in the determination target text, Kiyomizu-dera in Kyoto city where the determination target specific point name corresponds to the famous specific point It is determined that

  By performing such a determination process, the determination unit 18A makes an appropriate determination in consideration of the trend, for example, when the second-ranked TW number shows a sudden rise due to an event (for example, a festival). It can be carried out. Moreover, since the same-name specific point (same name specific point whose famousness is less than or equal to a predetermined value) that is unlikely to be applicable is excluded in advance, the specific point name can be extracted efficiently and accurately.

  Here, when the same name specific point indicated by the determination target specific point name could not be specified in the above-mentioned same name determination process (for any same name specific point, the corresponding surrounding place name or feature word is not included in the determination target text In this case, it may be determined that the determination target specific point name indicates a famous specific point. Thereby, the reproduction rate of extraction of a specific point name can be raised.

  Next, personal name / general word flag processing will be described. If the personal name flag or the general word flag (first flag) corresponding to the determination target specific spot name is set, the determination unit 18A is associated with the determination target specific spot name and stored in the feature word storage unit 15A. It is determined whether or not the word is included in the determination target text. When the feature word is included in the determination target text, the determination unit 18A determines that the determination target specific point name in the determination target text means a specific point. On the other hand, when the feature word is not included in the determination target text, the determination unit 18A determines that the determination target specific spot name in the determination target text means a person name or a general word.

  The determination unit 18A outputs determination result data in which a determination target text and a determination result are associated with each other. Here, the determination result is information indicating that the determination target text is determined to mean a specific point in the personal name / general word flag processing. The determination result data output by the determination unit 18A is stored in the determination result storage unit 19 that is a database or the like. This determination result data is accumulated for generation of feature words by a feature word generation unit 20A described later.

  As shown in FIG. 4, the feature word generation unit 20A generates a feature word based on the determination result data stored in the determination result storage unit 19, and stores the generated feature word in the feature word storage unit 15A. It is a generation means. The feature word generation unit 20A generates a feature word based on the determination result data stored in the determination result storage unit 19 within a predetermined period (for example, several hours to several days). Specifically, the feature word generation unit 20 extracts a specific expression from a determination target text included in the determination result data by a known specific expression extraction technique, for example, and the appearance frequency of the extracted specific expressions is equal to or higher than a predetermined threshold. A proper expression (related word) can be generated as a feature word. The feature word generation unit 20A may generate a feature word by using tfidf, a dice coefficient, or the like for the determination target text included in the determination result data. In the following description, the feature word generated based on the determination result data is also referred to as a dynamic feature word in distinction from the static feature word.

  The dynamic feature words generated by the feature word generation unit 20A are stored in the feature word storage unit 15A together with the static feature words stored in advance in the feature word storage unit 15A. The static feature words and dynamic feature words stored in the feature word storage unit 15A are used without being particularly distinguished in the determination by the determination unit 18A.

  In this way, the dynamic feature words generated based on the determination result data by the feature word generation unit 20A are stored as feature words in the feature word storage unit 15A, and thus the feature words stored in the feature word storage unit 15A are It will gradually increase. Thereby, the reproduction rate of extraction of the specific spot name in the determination by the determination unit 18A is increased. The frequency (cycle) at which the feature word generation unit 20 generates a dynamic feature word from the determination result data, the target period and the number of data of the determination result data used for generating the dynamic feature word, and the like are determined by, for example, an administrator Set arbitrarily.

  Next, operation | movement of 1 A of specific spot name determination apparatuses containing the specific spot name determination method which concerns on this embodiment is demonstrated using FIGS.

  As shown in FIG. 6, the text extraction unit 14 reads the specific spot information stored in the specific spot information storage unit 13 and loads the read specific spot information on the RAM 102 (step S <b> 101). The posted text is acquired from the text input unit 2 by the text acquisition unit 11, and the acquired posted text is stored in the text storage unit 12 (step S102). Subsequently, the text extraction unit 14 extracts the determination target text including the specific point name associated with the specific point information loaded on the RAM 102 from the plurality of posted texts stored in the text storage unit 12. (Step S103, text acquisition step).

  Subsequently, processing (step S104 to step S107) including a determination step of the specific spot name determination method is executed by the determination unit 18A. However, the feature word generation processing shown in step S304 included in step S107 is executed by the feature word generation unit 20A. First, it is determined whether or not the same name flag corresponding to the determination target specific point name is set (step S104). If the same name flag is set (step S104: YES), the same name flag process is executed (step S105). On the other hand, when the same name flag is not set (step S104: NO), it is determined whether or not a person name flag or a general word flag corresponding to the specific point name is set (step S106). If the personal name flag or the general word flag is set (step S106: YES), the personal name / general word flag process is executed (step S107). On the other hand, when neither the personal name flag nor the general language flag stands (step S106: NO), the determination target specific spot name is not ambiguous (the specific spot indicated by the determination target specific spot name is uniquely determined). The process is terminated.

  FIG. 7 is a flowchart of the same name flag processing shown in FIG. As shown in FIG. 7, in the same name flag process, first, the determination unit 18A refers to the famousness storage unit 17 to refer to the famousity of the same name specific point having the same name as the determination target specific point name (step) S201). Subsequently, the determination unit 18A determines whether or not the famous point of the specific point having the maximum famousness among the specific points with the same name (famous specific point) is greater than or equal to a predetermined threshold (step S202). When the famous degree of the famous specific point is greater than or equal to a predetermined threshold (step S202: YES), it is determined that the specific point name indicates the famous specific point (step S203). On the other hand, when the famous degree of the famous specific point is less than the predetermined threshold (step S202: NO), the same name determination process described above is executed (step S204).

  FIG. 8 is a flowchart showing the personal name / general word flag processing shown in FIG. As shown in FIG. 8, in the personal name / general word flag process, first, a feature word (static feature word) stored in the feature word storage unit 15A in association with the determination target specific spot name is included in the determination target text. Whether or not a combination of a specific point name and a feature word is included in the determination target text (step S301). When the characteristic word corresponding to the determination target specific point name is included in the determination target text (step S301: YES), the determination unit 18A means the determination target specific point name in the determination target text means the specific point. It is determined (points to a place name / facility name) (step S302). The determination result data associating the determination result with the determination target text is stored in the determination result storage unit 19 (step S303). Then, the feature word generation unit 20A generates a dynamic feature word based on the determination result data stored in the determination result storage unit 19 at an arbitrary timing predetermined by an administrator or the like, and the generated dynamic feature The words are stored in the feature word storage unit 15A (step S304). Thereby, when performing the determination process using the specific point name determination apparatus 1A next time, in step S301, the feature word generation unit 20A in step S304 together with the static feature words stored in advance by the administrator or the like. A determination process using the generated dynamic feature word is executed.

  On the other hand, when the characteristic word corresponding to the determination target specific spot name is not included in the determination target text (step S301: NO), the determination unit 18A determines that the determination target specific spot name is a person name or general in the determination target text. It is determined to mean a word (step S305).

  In the specific point name determination apparatus 1A described above, the number of feature words stored in the feature word storage unit 15A increases (dynamic feature words are added) by repeating a series of determination processes. In the determination by the unit 18A, it is possible to increase the possibility that the specific point name that means the specific point in the determination target text can be appropriately extracted (determined to mean the specific point). That is, it is possible to increase the recall of the extraction of the specific point name used as meaning the specific point in the text.

  Moreover, in this specific point name determination apparatus 1A, when there are a plurality of the same name specific points having the same name as the determination target specific point name (when the second flag is set), the determination unit 18A If the famousness is greater than or equal to a predetermined threshold, it can be determined that the determination target specific point name indicates the famous specific point without executing the above-described same name determination process. As a result, even when the surrounding text corresponding to the specific spot name is not included in the determination target text, the specific spot name can be appropriately extracted based on the famousness of the specific spot. The rate is improved.

  Moreover, in this specific point name determination apparatus 1A, the famousness calculation unit 16 appropriately calculates the famousity for each specific point with the same name based on the number of surrounding place name texts, the number of feature word texts, the number of check-ins, and the like. can do. And the determination part 18A determines the specific spot name based on the famousity calculated in this way, so that the accuracy of extraction of the specific spot name can be improved.

[Second Embodiment]
FIG. 9 is a block diagram showing a functional configuration of the specific point name determination apparatus 1B according to the second embodiment of the present invention. In the specific point name determination device 1B, the feature word storage unit 15A, the determination unit 18A, and the feature word generation unit 20A of the specific point name determination device 1A are the feature word storage unit 15B, the determination unit 18B, and the feature word generation unit 20B, respectively. Is different from the specific point name determination apparatus 1A. Further, the specific point name determination device 1B is different from the specific point name determination device 1A in that it includes a determination model storage unit 21. Hereinafter, the point that the specific point name determination apparatus 1B is different from the specific point name determination apparatus 1A will be mainly described.

  The feature word storage unit 15B is a feature word storage unit that stores an incorrect answer feature word, which is a word characterizing that a specific spot name does not mean a specific spot in the text, together with the feature word for each specific spot name. In the present embodiment, an incorrect answer characteristic word is a word that characterizes that a specific point name means a person name or general word other than the specific point in the text. Here, an incorrect answer feature word is an administrator or the like in advance based on one or more sentences in which the specific point name does not mean a specific point in the text (means a person name or general word in the text) And is stored in the feature word storage unit 15B. The incorrect answer feature word is extracted by, for example, tfidf or the like, like the feature word. Since the feature words stored in the feature word storage unit 15B are the same as the feature words stored in the feature word storage unit 15A, description thereof is omitted.

  The determination model storage unit 21 is a determination model storage unit that inputs a determination target text and stores a determination model for determining whether or not a specific point name means a specific point in the determination target text. The determination model is a learning model generated by machine learning, for example. This learning model is prepared in advance by an administrator or the like, for example. Here, as the machine learning, for example, a technique such as CRF (Conditional Random Fields) or SVM (Support Vector Machine) can be used. In addition, the learning model uses, for example, a text including a specific point name as an explanatory variable, and what the specific point name means in the text (used in any sense among a specific point, a person name, a general word, etc.) It is obtained by performing machine learning such as CRF and SVM based on supervised data with information indicating whether the object is a target variable.

  Here, the determination model is, for example, a model (person name model) for determining whether or not the specific point name means a person name, and a model (general language model) for determining whether or not the specific point name means a general word. As described above, the determination model may be configured by category.

  The determination unit 18B has a determination target specific point name in the determination target text based on the determination model stored in the determination model storage unit 21 and the feature words and incorrect answer feature words stored in the feature word storage unit 15B. It is a determination means for determining whether or not a specific point is meant.

  Specifically, when the determination unit 18B determines that the determination target specific point name does not mean the specific point in the determination target text based on the determination model, the determination unit 18B associates the determination target specific point name with the determination target specific point name. If the characteristic word memorize | stored in 15B is contained in the said determination target text, it will determine with the said determination target specific point name meaning the specific point in the said determination target text. On the other hand, when the determination unit 18B determines that the determination target specific spot name does not mean the specific spot in the determination target text based on the determination model, the determination unit 18B stores the determination target specific spot name in association with the determination target specific spot name in the feature word storage unit 15B. If the determined feature word is not included in the determination target text, it is determined that the determination target specific point name does not mean a specific point in the determination target text (means a person name or general term other than the specific point) To do.

  Further, when the determination unit 18B determines that the determination target specific spot name means a specific spot in the determination target text based on the determination model, the determination unit 18B is associated with the determination target specific spot name and stored in the feature word storage unit 15B. If the incorrect characteristic word is not included in the determination target text, it is determined that the determination target specific point name means the specific point in the determination target text. On the other hand, when the determination unit 18B determines that the determination target specific point name means a specific point in the determination target text based on the determination model, the determination unit 18B is associated with the determination target specific point name and stored in the feature word storage unit 15B. If the incorrect target characteristic word is included in the determination target text, the determination target specific point name in the determination target text does not mean a specific point (means a person name or general term other than the specific point) judge.

  The determination unit 18B outputs determination result data in which a determination target text and a determination result are associated with each other. Here, the determination result is information indicating whether the specific point name included in the determination target text is determined by the determination unit 18B to mean a specific point, a person name, or a general word. The determination result data output by the determination unit 18B is stored in the determination result storage unit 19. This determination result data is accumulated for generation of feature words or incorrect feature words by a feature word generation unit 20A described later.

  Here, in the case where determination models for each category (personal name model, general language model) are stored in the determination model storage unit 21, the determination unit 18B determines the determination target specific point name at the time of determination based on the determination model. The determination can be made using the determination model corresponding to the first flag (person name flag, general word flag) corresponding to. Specifically, when the personal name flag is set, the determination unit 18B determines the determination target text using the personal name model, so that the determination target specific point name in the determination target text means the personal name. It can be determined whether or not. When the general word flag is set, the determination unit 18B determines the determination target text using the general word model, and thus the determination target specific point name in the determination target text means a general word. It can be determined whether or not. When both the personal name flag and the general word flag are set, the determination unit 18B determines the determination target text using both the personal name model and the general word model, thereby determining the determination target text. It can be determined whether the target specific point name means a personal name or a general term.

  The feature word generation unit 20B generates a feature word or an incorrect answer feature word based on the determination result data stored in the determination result storage unit 19, and stores the generated feature word or incorrect answer feature word in the feature word storage unit 15B. Feature word generating means. Specifically, the feature word generation unit 20B extracts the determination target text including the specific point name determined to mean the specific point by the determination unit 18B from the determination result data, and extracts the characteristic word from the determination target text. Generate. In addition, the feature word generation unit 20B extracts the determination target text including the specific point name determined to mean a human name or a general word by the determination unit 18B from the determination result data, and the incorrect answer feature word from the determination target text Is generated. The feature word generation unit 20B can generate, for example, a proper expression (related word) extracted by a known proper expression extraction technique as a feature word or an incorrect answer feature word. The feature word generation unit 20B may generate a feature word or an incorrect answer feature word using tfidf, a dice coefficient, or the like.

  The feature word or the incorrect answer feature word generated by the feature word generation unit 20B is stored in the feature word storage unit 15B together with the feature word and the incorrect answer feature word stored in the feature word storage unit 15B in advance. A feature word or an incorrect answer feature word stored in the feature word storage unit 15B by the feature word generation unit 20B and a feature word and an incorrect answer feature word stored in advance in the feature word storage unit 15B by an administrator or the like are determined. The determination by the unit 18A is used without being particularly distinguished.

  Next, operation | movement of the specific point name determination apparatus 1B is demonstrated using FIG. First, the determination unit 18B determines whether the determination target specific spot name means a specific spot in the determination target text based on the determination model (step S401). When it is determined in step S401 that the determination target specific spot name does not mean a specific spot in the determination target text (step S401: NO), the determination target specific spot name is stored in the feature word storage unit 15B in association with the determination target specific spot name. It is determined whether or not the feature word is included in the determination target text (step S402). On the other hand, when it is determined in step S401 that the determination target specific spot name means a specific spot in the determination target text (step S401: YES), the determination target specific spot name is stored in the feature word storage unit 15B in association with the determination target specific spot name. It is determined whether or not the incorrect feature word that has been included is included in the determination target text (step S403).

  If the feature word associated with the determination target specific spot name is not included in the determination target text in step S402 (step S402: NO), the determination target specific spot name does not mean a specific spot in the determination target text (specific (It refers to a person name or general term other than the point)) (step S405). On the other hand, if the determination target text includes a feature word associated with the determination target specific spot name in step S402 (step S402: YES), the determination target specific spot name in the determination target text means a specific spot. It is determined (points to a place name / facility name) (step S404).

  If the incorrect characteristic word associated with the determination target specific point name is not included in the determination target text in step S403 (step S403: NO), the determination target specific point name means the specific point in the determination target text. It is determined (points to a place name / facility name) (step S404). On the other hand, if the incorrect target characteristic word associated with the determination target specific spot name is included in the determination target text in step S403 (step S403: YES), the determination target specific spot name in the determination target text indicates the specific spot. It is determined that it does not mean (refers to a person name or general term other than a specific point) (step S405).

  Subsequently, determination result data in which the determination result in step S404 or step S405 is associated with the determination target text is stored in the determination result storage unit 19 (step S406). Then, the feature word generation unit 20B generates a dynamic feature word or an incorrect feature word based on the determination result data stored in the determination result storage unit 19 at an arbitrary timing predetermined by an administrator or the like, The generated dynamic feature word or incorrect feature word is stored in the feature word storage unit 15B (step S407). Thereby, when performing the determination process using the specific point name determination apparatus 1B next time, it is generated by the feature word generation unit 20B in step S407 together with the feature words stored in advance by the administrator or the like in step S402. The determination process using the feature word is executed. In step S403, the determination process using the incorrect feature word generated by the feature word generation unit 20B in step S407 together with the incorrect feature word previously stored by the administrator or the like is executed.

  According to the specific point name determination apparatus 1B described above, the determination by the determination model and the determination using the feature word or the incorrect answer feature word are performed together to use the specific point in the text. It is possible to increase the reproducibility of the extraction of the name of the specific spot and to increase the accuracy (matching rate). Specifically, even when the determination unit 18B determines that the specific point name does not mean the specific point in the determination target text by the determination by the determination model, the specific point name means the specific point. Is included in the determination target text, it is determined that the specific point name means a specific point. By such determination, the reproduction rate of the extraction of the specific point name is increased. On the other hand, if the determination unit 18B determines that the specific point name means a specific point in the determination target text by the determination by the determination model, the incorrect answer feature word characterizing that the specific point name does not mean the specific point Is not included, it is determined that the specific point name in the determination target text means a specific point. Such determination can improve the accuracy of extraction of the specific spot name.

  Next, the specific point name determination program P1 according to one embodiment of the present invention will be described using FIG. The specific point name determination program P1 is a program for causing a computer to function as the specific point name determination device 1A.

  FIG. 11 is a block diagram showing modules of the specific point name determination program P1. As shown in FIG. 11, the specific point name determination program P1 includes a specific point information storage module P11, a feature word storage module P12, a text acquisition module P13, a text extraction module P14, a determination module P15, and a feature word generation. Module P16. The functions realized by executing the specific point information storage module P11, the feature word storage module P12, the text acquisition module P13, the text extraction module P14, the determination module P15, and the feature word generation module P16 are as described above. The functions of the specific point information storage unit 13, the feature word storage unit 15A, the text acquisition unit 11, the text extraction unit 14, the determination unit 18A, and the feature word generation unit 20A corresponding to the spot name determination device 1A are the same.

  The specific point name determination program P1 configured as described above is stored in a recording medium such as a CD-ROM and a DVD, and is executed by a computer used as the specific point name determination device 1A. Specifically, the computer includes a recording medium reading unit such as a CD-ROM drive and a DVD drive. When the recording medium is inserted into the recording medium reading unit, the computer can access the specific point name determination program P1 stored in the recording medium from the recording medium reading unit. Then, by causing the computer to execute the specific point name determination program P1, the computer can be operated as the specific point name determination device 1A.

  The specific point name determination program P1 may be provided via a network as a data signal superimposed on a carrier wave. In this case, the computer used as the specific location name determination apparatus 1A can execute the specific location name determination program P1 by storing the specific location name determination program P1 received by the communication module 106 in the CPU 101 or the RAM 102.

  The present invention has been described in detail based on the embodiments. However, the present invention is not limited to the above embodiment. The present invention can be variously modified without departing from the gist thereof.

  For example, in this embodiment, POIs of facilities (Kiyomizu-dera), lakes (Oonuma), and mountains (Mt. Hakodate) are shown as examples of specific points, but the specific points are, for example, sightseeing spots, downtown areas, station fronts, etc. Anything can be used as long as it indicates a point.

  Moreover, in this embodiment, although a person name and a general term were mentioned as an example of what means other than a specific point, you may use another category. For example, when it is desired to exclude a station name from a specific point, a station name flag indicating the same name as the station name may be used as the first flag.

  Further, in the present embodiment, it has been described that either the same name flag process or the personal name / general word flag process is performed depending on the case according to the state of the flag, but both flags are set. Alternatively, both processes may be executed to make a comprehensive determination. In the present embodiment, priority is given to the same name flag processing, but personal name / general word flag processing may be prioritized.

  DESCRIPTION OF SYMBOLS 1A, 1B ... Specific point name determination apparatus, 2 ... Text input part, 11 ... Text acquisition part, 12 ... Text storage part, 13 ... Specific point information storage part, 14 ... Text extraction part, 15 ... Feature word storage part, 16 ... famousity calculation unit, 17 ... famousity calculation unit, 18A, 18B ... determination unit, 19 ... determination result storage unit, 20A, 20B ... feature word generation unit, 21 ... determination model storage unit, 101 ... CPU, 102 ... RAM , 103 ... ROM, 104 ... input device, 105 ... output device, 106 ... communication module, 107 ... auxiliary storage device, P1 ... specific location name determination program, P11 ... specific location information storage module, P12 ... feature word storage module, P13 ... text acquisition module, P14 ... determination module, P15 ... feature word generation module.

Claims (8)

Specific point information storage means for storing specific point information that associates a specific point name indicating a name of a predetermined specific point with a first flag indicating that the specific point name has a meaning other than the specific point When,
Feature word storage means for storing, for each specific point, a characteristic word that is a word characterizing that the specific point name means the specific point in the text;
A text acquisition means for acquiring text;
A text extraction unit that extracts a determination target text including a specific point name associated with the specific point information from the text acquired by the text acquisition unit;
The first flag corresponding to the specific spot name included in the determination target text indicates that the specific spot name has a meaning other than the specific spot, and the feature word storage means is associated with the specific spot name. If the feature word stored in the determination target text is included in the determination target text, determination means for determining that the specific point name means the specific point in the determination target text;
Feature word generating means for generating the feature word based on determination result data in which the determination target text is associated with a determination result by the determination means for the determination target text, and storing the generated feature word in the feature word storage means When,
A specific point name determination device comprising: A determination model storage means for storing a determination model for inputting the determination target text and determining whether the specific point name means the specific point in the determination target text;
The feature word storage means stores an incorrect answer feature word that is a word characterizing that the specific point name does not mean the specific point in the text together with the characteristic word for each specific point name,
When the determination unit determines that the specific point name included in the determination target text does not mean the specific point in the determination target text based on the determination model, the characteristic word is associated with the specific point name. If the feature word stored in the storage means is included in the determination target text, it is determined that the specific point name means the specific point in the determination target text, and the determination target is based on the determination model. When it is determined that the specific point name included in the text means the specific point in the determination target text, an incorrect feature word stored in the characteristic word storage unit in association with the specific point name is the determination target If it is not included in the text, it is determined that the specific point name means the specific point in the determination target text.
The feature word generation means generates the feature word or the incorrect answer feature word based on the determination result data, and stores the generated feature word or incorrect answer feature word in the feature word storage means.
The specific point name determination apparatus according to claim 1. The specific point information is associated with a second flag indicating that there are a plurality of the same name specific points which are specific points having the same name as the specific point name associated with the specific point information,
For each of the same-name specific points, further comprising a famousness storage means for storing a pre-calculated famousness,
If the second flag corresponding to the specific point name included in the determination target text indicates that there are a plurality of the same name specific points, the determination unit is stored in the famousness storage unit. Determine whether to execute the same name determination process based on the famousness of the same name specific point
Based on a character string other than the specific point name included in the determination target text, the same name determination process indicates which specific point among the same name specific points in the determination target text. Is a process for determining
The specific point name determination apparatus according to claim 1 or 2. The determination means does not perform the same name determination process when the famousness of the specific point having the highest famousness among the same name specific points is equal to or greater than a predetermined threshold, and the specific point included in the determination target text It is determined that the point name indicates the specific point having the maximum famousness, and the same name determination process is executed when the famous point of the specific point having the maximum famousness among the specific points having the same name is less than a predetermined threshold To
The specific point name determination apparatus according to claim 3. The determination means, when the famousness of the specific point having the highest famousness among the same name specific points is less than a predetermined threshold, out of the same name specific points, the same name specific point having the famousness below a predetermined value. Exclude and execute the same name determination process,
The specific point name determination apparatus according to claim 4. The number of surrounding place name texts including both the specified point and the surrounding place name corresponding to the specified point, the number of feature word texts including both the specified point and the feature word corresponding to the specified point, and Calculating the famousness of the specific point based on at least one of the check-in numbers indicating the number of times the user of the mobile communication terminal has stayed at the specific point, and storing the calculated famousness in the famousity storage means Further comprising a celebrity calculation means,
The specific point name determination apparatus as described in any one of Claims 3-5. Specific point information storage means for storing specific point information that associates a specific point name indicating a name of a predetermined specific point with a first flag indicating that the specific point name has a meaning other than the specific point When,
Feature word storage means for storing, for each specific point, a characteristic word that is a word characterizing that the specific point name means the specific point in the text;
A specific point name determination method executed by a specific point name determination device comprising:
A text acquisition step for acquiring text;
A text extraction step for extracting a determination target text including a specific point name associated with the specific point information from the text acquired in the text acquisition step;
The first flag corresponding to the specific spot name included in the determination target text indicates that the specific spot name has a meaning other than the specific spot, and the feature word storage means is associated with the specific spot name. If the characteristic word stored in the determination target text is included in the determination target text, a determination step for determining that the specific point name means the specific point in the determination target text;
A feature word generation step of generating the feature word based on determination result data in which the determination target text is associated with a determination result in the determination step for the determination target text, and storing the generated feature word in the feature word storage unit When,
Specific point name determination method including Computer
Specific point information storage means for storing specific point information that associates a specific point name indicating a name of a predetermined specific point with a first flag indicating that the specific point name has a meaning other than the specific point When,
Feature word storage means for storing, for each specific point, a characteristic word that is a word characterizing that the specific point name means the specific point in the text;
A text acquisition means for acquiring text;
A text extraction unit that extracts a determination target text including a specific point name associated with the specific point information from the text acquired by the text acquisition unit;
The first flag corresponding to the specific spot name included in the determination target text indicates that the specific spot name has a meaning other than the specific spot, and the feature word storage means is associated with the specific spot name. If the feature word stored in the determination target text is included in the determination target text, determination means for determining that the specific point name means the specific point in the determination target text;
Feature word generating means for generating the feature word based on determination result data in which the determination target text is associated with a determination result by the determination means for the determination target text, and storing the generated feature word in the feature word storage means Specific point name determination program to function as.

Images