JP2011118797A - Mobile cgm analysis system, method, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a mobile CGM (Consumer Generated Media) analysis system for analyzing reputation for each attribute, based on a text file including a pictogram posted from a cellular phone and attribute information of a user. <P>SOLUTION: The CGM analysis system analyzes an evaluation level in a text file posted from a portable terminal, and adds an attribute of a user who has posted a text file. The text file includes character strings and pictograms and is associated with a user identifier, and based on the user identifier, the attribute of the user is associated with the text file. An evaluation pictogram which is a specific pictogram associated with an evaluation level is registered. The text file is decomposed into morphemes, and a morpheme corresponding to the evaluation pictogram is extracted using a pictogram dictionary. Using the pictogram dictionary, an evaluation level for the extracted evaluation pictogram is retrieved. Based on the sum of the evaluation levels of all the evaluation pictograms, an evaluation level of the text file is decided. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a system, method, and program for analyzing CGM data read and written by a mobile terminal such as a mobile phone.

  Information obtained by evaluating or commenting on a product or service is often transmitted by a seller or an advertising agency. On the other hand, on the Internet site, the actual experiences and live voices of consumers themselves have been accumulated in a huge amount in real time. Information that is not bound by such interests can be extremely important support information for other consumers.

  As media generated by many general consumers, CGM (Consumer Generated Media) using the Internet has attracted attention. Specifically, it is a Web site in which information sent / posted by individuals is made into a database and media, and is one of Web 2.0. For example, there are a blog (Web log), a review site, a Q & A community, an SNS (Social Networking Service), and a COI (Community Of Interest) site. The information exchanged here is not limited to information about products and services, but includes various information such as the contents of daily events.

  However, the amount of information of such CGM is extremely large, and the information content is complicated. The information evaluated or commented on by an individual is generally a text file including a character string. Therefore, the third party who browses needs to read the contents of the sentence file, and cannot grasp the evaluation contents at a glance. Moreover, it becomes more difficult to analyze the evaluation of many general consumers.

  On the other hand, in recent years, a “reputation analysis service” that analyzes a reputation by searching for text files sent and posted by individuals on the Internet site and scoring the contents has been provided (for example, non-reputation service). (See Patent Documents 1 and 2). The reputation analysis server that executes the service first reads a text file and performs morphological analysis on the text file. And the meaning of a morpheme is scored, and the reputation with respect to the sentence file is scored (for example, refer patent document 1). A third party who browses can view the intended reputation of the text file simply by looking at the reputation score.

Japanese Patent No. 3962382

"Nikkei Research blogViz Sensor", Nikkei Research, Inc.), [online], [Searched on November 6, 2009], Internet <URL: http://viz.nikkei-r.co.jp/> “Dentsu Buzz Research”, Dentsu Hotlink, [online], [searched on November 6, 2009], Internet <URL: https://www.dbuzz.jp/>

  In recent years, with the development of mobile communication environments, text files are often sent and posted from mobile phones to Web servers. At this time, the text posted from the mobile phone is different from the text posted from the personal computer and often does not conform to the normal Japanese grammar. Specifically, in mobile phone applications, “pictograms” are often used as emotional expressions in text files. The user uses this pictogram to supplement and emphasize words expressing emotions.

  However, when pictograms are inserted into the text file, the existing reputation analysis service does not recognize the text as a normal text, and the dependency analysis fails. For example, there are cases where pictograms are originally inserted where the subject should come. Moreover, even if the emotional expression for a certain character string is emphasized or supplemented, the pictogram itself is recognized only as a symbol. After all, even though the user of the mobile phone uses pictograms as emotional expressions themselves, the reputation analysis service cannot perform reputation analysis based on the pictograms.

  Further, a document transmitted / posted to CGM on the Internet cannot specify the attributes of the sender / poster. The available attributes were limited to gender analysis at best. In addition, since it is an assumed value determined from the text content of a blog or the like, the accuracy is low. Therefore, it was difficult to determine whether the results analyzed by existing reputation analysis services were applicable regardless of all generations and genders. Therefore, it could not be used for detailed marketing by gender and generation.

  In addition, writing from mobile phones tends to be performed by young people who do not have a personal computer. As a result, it is expected that many products will be evaluated by young people, but this cannot be supported by existing reputation analysis services.

  Accordingly, an object of the present invention is to provide a mobile CGM analysis system, method and program capable of performing reputation analysis for each attribute from a text file including pictographs posted from a mobile phone and user attribute information.

  In order to achieve the above object, a mobile CGM analysis system according to the present invention is a mobile CGM analysis system that analyzes an evaluation level in a text file posted from a mobile terminal and assigns an attribute of a text file submitter. Includes a character string and a pictogram, and is further associated with a contributor identifier. Based on the contributor identifier, attribute information providing means for associating the attribute of the contributor with the sentence file, and corresponding evaluation level to the specific pictogram A pictogram dictionary means for registering the attached evaluation pictogram, a morpheme analysis means for decomposing the sentence file into morphemes, a pictogram extraction means for extracting the morpheme corresponding to the evaluation pictogram using the pictogram dictionary means, and Using the pictogram dictionary means, search the evaluation level for the extracted evaluation pictogram And pictographs level search means that, based on the sum of the evaluation pictogram all evaluation levels, and an evaluation level determining means for determining an evaluation level of the document file.

  It is also preferable to further include means for inputting contractor information of the mobile terminal, and the attribute information giving means obtains the attribute of the contributor of the text file from the contractor information and the contributor identifier. .

  Further, the server further includes means for inputting a log on the server, and a log information adding unit that associates the log related to the document file with the document file, the identification information being equal to the contributor identifier of the document file, It is also preferable.

  Moreover, it is also preferable that the attribute is the sex, age, and residential area of the person who posted the text file.

  In order to achieve the above object, a mobile CGM analysis method according to the present invention is a mobile CGM analysis method for analyzing an evaluation level in a text file posted from a mobile terminal and adding an attribute of a text file submitter. Includes a character string and a pictogram, is further associated with a poster identifier, has a pictogram dictionary for registering a rating pictogram in which an evaluation level is associated with a specific pictogram, and based on the poster identifier, Attribute information giving step for associating an attribute with the sentence file, a morpheme analyzing step for decomposing the sentence file into morphemes, a pictogram extracting step for extracting the morpheme corresponding to the evaluation pictogram using the pictogram dictionary, Search the evaluation level for the extracted evaluation pictogram using a pictogram dictionary A character-level search step, on the basis of the sum of the evaluation pictogram all evaluation levels, and an evaluation level determination step of determining the evaluation level of the document file.

  In order to achieve the above object, the program according to the present invention is a program for analyzing an evaluation level in a text file posted from a mobile terminal, and causing a computer mounted on a server to give an attribute of a text file poster, The sentence file includes a character string and a pictogram, and is further associated with a contributor identifier. Based on the contributor identifier, attribute information giving means for associating the attribute of the contributor with the sentence file, and evaluating the specific pictogram Pictogram dictionary means for registering evaluation pictograms associated with levels, morpheme analysis means for decomposing the text file into morphemes, and pictogram extraction means for extracting the morphemes corresponding to the evaluation pictograms using the pictogram dictionary means And using the pictogram dictionary means, the evaluation label for the evaluation pictogram extracted is used. A pictogram level search means for searching the Le, on the basis of the sum of the evaluation pictogram all evaluation levels, causing a computer to function as an evaluation level determining means for determining the evaluation level of the document file.

  As described above, according to the mobile CGM analysis system, method, and program of the present invention, it is possible to perform reputation analysis for each attribute from a text file including pictographs posted from a mobile phone and user attribute information. Detailed marketing data can be provided.

It is a system configuration diagram having a mobile CGM analysis system. An example of identification of a contributor of a text file on a word-of-mouth site is shown. It is a functional block diagram of the mobile CGM analysis system in this invention. It is a character string dictionary table showing evaluation with respect to a specific character string. It is a pictogram dictionary table showing evaluation with respect to a specific pictogram and adverbial magnification. It is a system block diagram which has a mobile CGM analysis system in other embodiment of this invention. It is a flowchart in this invention.

  Hereinafter, the best mode for carrying out the present invention will be described in detail with reference to the drawings. In the present embodiment, a document file posted on a word-of-mouth site will be described, but the same applies to blogs, Q & A communities, and SNSs.

  FIG. 1 is a system configuration diagram having a mobile CGM analysis system.

  According to FIG. 1, a plurality of terminals and a mobile phone 3 can send and post to a word-of-mouth site 4 via a mobile phone network. The word-of-mouth site 4 accumulates the text files transmitted / posted from the mobile phone 3 and discloses them to a third party. The mobile CGM analysis system 1 takes in and stores text files from the word-of-mouth site 4 and contractor information from the mobile phone carrier 2, and assigns attributes to the text files to analyze the evaluation level. In the present embodiment, the text file and the contractor information are obtained by the mobile CGM analysis system 1 via the mobile phone network, but can also be obtained via the Internet or a dedicated line. Furthermore, it is possible to write to an external storage medium such as a magnetic recording medium or a flash memory, and the mobile CGM analysis system 1 can read from the external storage medium.

  The “sentence file” in the present invention further includes not only a character string as text but also “pictograms” different from existing text. “Emoji” refers to a simplified picture used to convey meaning (eg, emotional expression). This is also called a pictogram, for example, and anyone (for example, a foreigner who cannot understand the language) can understand the meaning of the display regardless of the existing language system.

  Pictographs are used especially for sending and receiving mail between mobile phones. For example, various icons such as “smile”, “crying face”, “heart mark” for expressing emotions are prepared. Pictographs are used as icons for easily expressing emotions when writing mail containing text.

  “Emoji” has a character code different from that of general text. Usually, the text means ASCII code such as single-byte alphanumeric characters, JIS code such as Japanese Kanji or Shift JIS, Unicode for multiple languages, and the like. On the other hand, each mobile phone communication carrier uniquely determines a pictogram and its character code so that the pictogram does not overlap with these character codes. Therefore, when the telecommunications carrier or the mobile phone manufacturer is different, it is not always possible to display the same pictograph. In addition, a specific communication carrier may transmit pictogram contents after an escape code. This escape code means that “the subsequent data is a pictograph”. Therefore, a pictograph cannot be displayed by a text editor as an application of a personal computer. Thus, text and pictograms have different character systems.

  The word-of-mouth site 4 of this system is installed in the mobile phone network, and the “text file” in the present invention is limited to the file posted from the mobile phone 3. Therefore, this document file can identify the contributor.

  FIG. 2 shows an example of identification of a contributor of a text file on a word-of-mouth site. The document file posted on the word-of-mouth site 4 has poster identification information attached thereto. The poster identification information is, for example, a mobile phone subscriber ID. The mobile CGM analysis system 1 receives the document file and the poster identification information from the word-of-mouth site 4. The document file and the poster identification information need not be received at the same time. It is also possible to receive the document file and the poster identification information in association with each other. The mobile CGM analysis system 1 receives contractor information from the mobile phone carrier 2. As shown in the example of FIG. 2, the contractor information is data owned by the mobile phone carrier 2 and includes information such as the mobile phone subscriber ID, gender, age, and region at the time of subscription. Have. The mobile CGM analysis system 1 can know the attributes (gender, age, assumed residence area, etc.) of the document file poster from the poster identification information and contractor information of the document file. The assumed residence area is the residence area of the document file contributor assumed from the area at the time of subscription.

  FIG. 3 is a functional configuration diagram of the mobile CGM analysis system according to the present invention.

  According to FIG. 3, the mobile CGM analysis system 1 includes a mobile phone network interface unit 101 that communicates with the word-of-mouth site 4 and the mobile phone carrier 2, and a mobile review that stores posted / viewed text files and contractor information. And a data storage unit 102.

  The mobile CGM analysis system 1 for the evaluation analysis service further includes an attribute information addition unit 103, a morpheme analysis unit 104, a character string dictionary unit 105, a character string extraction unit 106, a character string level search unit 107, It has a pictogram dictionary unit 108, a pictograph extraction unit 109, a pictograph level search unit 110, an adverbial magnification search unit 111, a dependency analysis unit 112, and an evaluation level determination unit 113. These functional components are realized by executing an evaluation analysis program that causes a computer mounted on the server to function.

  The attribute information adding unit 103 reads the text file from the mobile word-of-mouth data storage unit 102, associates the document file with the contractor information from the poster identification information and contractor information of the document file, and sets the attribute of the poster in the text file. Give. As attributes, for example, sex, age, and residential area are given.

  The morpheme analysis unit 104 reads the text file to which the attribute is added from the attribute information addition unit 103, and decomposes the text file into morphemes. The decomposed morphemes are output to the character string extraction unit 106, the pictogram extraction unit 109, and the adverbial magnification search unit 111.

  The character string dictionary unit 105 registers an evaluation character string in which an evaluation level is associated with a specific character string as shown in FIG. For example, the evaluation level may be divided into “positive” / “negative” in stages.

  According to the table in FIG. 4, the evaluation is associated with each specific character string. For example, “positive P” is associated with the character strings “special” and “excitement”. On the other hand, “Negative N” is associated with the character strings “disappointing” and “stupid”.

  The character string extraction unit 106 uses the character string dictionary unit 105 to extract morphemes corresponding to the evaluation character string. The evaluation character string is output to the character string level search unit 107.

  The character string level search unit 107 uses the character string dictionary unit 105 to search for an evaluation level for the extracted evaluation character string. The evaluation level is output to the evaluation level determination unit 113.

  As shown in FIG. 5, the pictogram dictionary unit 108 registers an evaluation pictogram in which an evaluation level is associated with a specific pictogram. Also, adverbic pictograms that associate adverbial magnification with specific pictograms are further registered.

  According to the table in FIG. 5, the category, title, evaluation, and adverbial magnification are associated with each specific pictogram. Here, “evaluation” is most simply associated with, for example, a positive / negative level. For example, points are assigned to positive “+1” and negative “−1”. Then, by summing the scores of all the pictograms in the text file, it is possible to express “how much is positive or negative” for the text file.

  According to FIG. 5, for example, “positive P” is associated with the pictograph of the title “Kirakira” (category “feelings / body”). Also, “positive P” is associated with the pictograph of the title “heart” (category “feelings / body”). On the other hand, “Negative N” is associated with the pictogram of the title “Teardrop” (category “face / expression”). Also, “negative N” is associated with the pictograph of the title “Aseri” (category “face / expression”).

  Further, according to FIG. 5, for example, an adverbial magnification of 1.2 is associated with a pictograph of the title “emphasis” (category “feelings / body”). Further, an adverbial magnification 1.1 is associated with the pictograph of the title “Winning Pride” (category “face / expression”).

  The pictograph extraction unit 109 uses the pictograph dictionary unit 108 to extract morphemes corresponding to the evaluation pictograph. The evaluation pictogram is output to the pictogram level search unit 110. The pictogram extraction unit 109 also uses the pictogram dictionary unit 108 to extract morphemes corresponding to adverbic pictograms. The adverbial pictogram is output to the adverbial magnification search unit 111.

  The pictogram level search unit 110 uses the pictogram dictionary unit 108 to search for an evaluation level for the extracted evaluation pictogram. The evaluation level is output to the evaluation level determination unit 113.

  The adverbial magnification search unit 111 uses the pictogram dictionary unit 108 to search for an adverbial magnification for the extracted adverbial pictograph. The adverbial magnification is output to the dependency analysis unit 112.

  The dependency analysis unit 112 detects an evaluation pictogram or an evaluation character string that is a dependency of an adverbial pictogram. This is determined by existing parsing techniques.

  The evaluation level determination unit 113 determines the evaluation level of the text file based on the sum of the evaluation levels of all character strings and evaluation pictograms. At this time, the evaluation level of the evaluation pictogram or evaluation character string that is a dependency of the adverbial pictogram is multiplied by the adverbial magnification of the adverbial pictogram. And the score for every morpheme in a text file is totaled. The evaluation level is associated with the sentence file in the mobile word-of-mouth data storage unit 102.

  As described above, the evaluation level and attributes are assigned to the posted text file. From this data, it is possible to grasp the evaluation for each user attribute of the product. For example, the user's evaluation (positive / negative) for the raw caramel of Company A can be grasped by sex, age, and region. Using this data, manufacturers can use it for effective advertising activities such as behavioral targeting advertising. For example, when the evaluation is low (negative) in a region where the raw caramel of Company A is present, fine marketing such as performing a different advertisement only for that region is possible.

  FIG. 6 is a system configuration diagram including a mobile CGM analysis system according to another embodiment of the present invention.

  According to FIG. 6, the system further includes a search site 5 and an EC site 6. The search site 5 is a site that provides a service for searching for goods or services related to the search word in the mobile phone network and the Internet by inputting the search word. The EC site 6 is a site that sells goods or services on the mobile phone network and the Internet.

  FIG. 6 shows a case where a user of the mobile phone 3 who has sent and posted a product evaluation document from the mobile phone 3 to the word-of-mouth site 4 searches and purchases product information on the mobile phone network in advance. The mobile CGM analysis system 1 of this embodiment obtains a search log from the search site 5 and a purchase log from the EC site 6. These pieces of information can be obtained online or offline like the contractor information. Similarly to the word-of-mouth site 4, the search site 5 and the EC site 6 are installed in the mobile phone network. Therefore, searcher identification information and purchaser identification information are included in the search log and purchase log. This identification information is also a mobile phone subscriber ID, for example. With this identification information, the search log and purchase log can be associated with the posting file of the word-of-mouth site 4.

  The mobile CGM analysis system 1 in this embodiment further includes a search log storage unit and a purchase log storage unit. The cellular phone network interface unit 101 can communicate with the search site 5 and the EC site 6. Further, the mobile CGM analysis system 1 includes a log information adding unit that associates a document file with a search log and a purchase log. The log information adding unit searches the search log and the purchase log for a log having the searcher identification information and the purchaser identification information equal to the poster identification information. Among them, the product log evaluated by the document file is further searched. The searched search log and purchase log are associated with the document file.

  As described above, in the present embodiment, the search log and the purchase log are associated with the document file in addition to the evaluation level and the attribute. Thereby, it is possible to grasp a series of flows of user consumption behavior (search-> purchase-> word-of-mouth) on the mobile phone. By using such data, companies can perform detailed marketing according to user consumption behavior and hobbies and preferences. It is possible to use it for behavioral targeting advertisement and the like by using the search behavior until a user who purchases a specific product reaches the product. Moreover, it can utilize for a recommendation service etc. using the attribute of the user who carries out positive evaluation of specific goods, and the purchase behavior of other goods.

  FIG. 7 is a flowchart in the present invention.

  FIG. 7 illustrates steps in an evaluation analysis program that causes a computer mounted on a server to be executed. The pictogram dictionary explained in FIG. 5, the character string dictionary explained in FIG. 4, and the contractor information are stored in advance. S104 is a step performed in the case of another embodiment.

(S101) A text file including a character string and a pictograph is read.
(S102) The text file is associated with the contractor information.
(S103) Assign attributes to the text file.
(S104) The text file is associated with the search log and the purchase log.
(S105) The sentence file is decomposed into morphemes.
(S106) The morpheme corresponding to the evaluation character string is extracted using the character string dictionary.
(S107) The evaluation level for the extracted evaluation character string is searched using the character string dictionary.
(S108) Extract morphemes corresponding to the evaluation pictograms using the pictogram dictionary.
(S109) The evaluation level for the extracted evaluation pictogram is searched using the pictogram dictionary.
(S110) Using the pictogram dictionary, the adverbial magnification with respect to the extracted adverb pictogram is searched.
(S111) An evaluation pictogram or evaluation character string that is a dependency of an adverbial pictogram is detected.
(S112) The evaluation level of the text file is determined based on the sum of the evaluation levels of all evaluation character strings and the evaluation levels of all evaluation pictograms. At this time, the evaluation level of the evaluation pictogram or the evaluation character string is multiplied by the adverbial magnification of the adverbial pictogram.
(S113) The evaluation level is associated with the sentence file.

  As described above in detail, according to the mobile CGM analysis server, method and program of the present invention, it is possible to perform reputation analysis for each attribute from a text file including pictograms posted from a mobile phone and user attribute information. It will be possible to provide more detailed marketing data. Furthermore, according to another embodiment, the search log and the purchase log are further associated, and fine marketing according to the user's consumption behavior / hobby preference becomes possible. Further, according to the above-described embodiment, the sentence file is described as being in Japanese, but reputation analysis is evaluated according to the present invention as long as pictorial characters are included in the sentence file, even in English and other languages. be able to.

  Moreover, all the embodiments described above are illustrative of the present invention and are not intended to limit the present invention, and the present invention can be implemented in other various modifications and changes. Therefore, the scope of the present invention is defined only by the claims and their equivalents.

DESCRIPTION OF SYMBOLS 1 Mobile CGM analysis system 2 Mobile phone communication provider 3 Mobile phone 4 Review site 5 Search site 6 EC site 101 Mobile phone network interface unit 102 Mobile word-of-mouth data storage unit 103 Attribute information addition unit 104 Morphological analysis unit 105 Character string dictionary unit 106 Character string extraction unit 107 Character string level search unit 108 Pictogram dictionary unit 109 Pictograph extraction unit 110 Pictogram level search unit 111 Adverbial magnification search unit 112 Dependency analysis unit 113 Evaluation level determination unit

Claims (6)

A mobile CGM analysis system for analyzing an evaluation level in a text file posted from a mobile terminal and adding an attribute of a text file submitter,
The sentence file includes a character string and a pictogram, and is further associated with a poster identifier,
Attribute information giving means for associating the attribute of the poster with the sentence file based on the poster identifier,
Pictogram dictionary means for registering an evaluation pictogram in which an evaluation level is associated with a specific pictogram,
Morphological analysis means for decomposing the sentence file into morphemes;
Pictogram extraction means for extracting the morpheme corresponding to the evaluation pictogram using the pictogram dictionary means;
Pictogram level search means for searching for the evaluation level for the extracted evaluation pictogram using the pictogram dictionary means;
An evaluation level determining means for determining an evaluation level of the sentence file based on a sum of evaluation levels of all the evaluation pictograms;
A mobile CGM analysis system comprising: A means for inputting contractor information of the mobile terminal;
2. The mobile CGM analysis system according to claim 1, wherein the attribute information adding unit obtains an attribute of the poster of the text file from the contractor information and the poster identifier. A means to enter logs on the server;
Among the logs, a log information giving unit that associates a log related to the document file with identification information equal to a contributor identifier of the document file,
The mobile CGM analysis system according to claim 1, further comprising:   The mobile CGM analysis system according to any one of claims 1 to 3, wherein the attributes are a sex, an age, and a residential area of the person who posted the text file. A mobile CGM analysis method for analyzing an evaluation level in a text file posted from a mobile terminal and adding an attribute of a text file submitter,
The sentence file includes a character string and a pictogram, and is further associated with a poster identifier,
Having a pictogram dictionary for registering an evaluation pictogram in which an evaluation level is associated with a specific pictogram,
Attribute information giving step for associating the attribute of the poster with the sentence file based on the poster identifier,
A morpheme analysis step for decomposing the sentence file into morphemes;
A pictogram extraction step of extracting the morpheme corresponding to the evaluation pictogram using the pictogram dictionary;
A pictogram level search step of searching for the evaluation level for the extracted evaluation pictogram using the pictogram dictionary;
An evaluation level determination step for determining an evaluation level of the sentence file based on a sum of evaluation levels of all the evaluation pictograms;
A mobile CGM analysis method characterized by comprising: A program that analyzes the evaluation level of a text file posted from a mobile terminal and functions a computer mounted on a server that gives the attribute of the text file submitter,
The sentence file includes a character string and a pictogram, and is further associated with a poster identifier,
Attribute information giving means for associating the attribute of the poster with the sentence file based on the poster identifier,
Pictogram dictionary means for registering an evaluation pictogram in which an evaluation level is associated with a specific pictogram,
Morphological analysis means for decomposing the sentence file into morphemes;
Pictogram extraction means for extracting the morpheme corresponding to the evaluation pictogram using the pictogram dictionary means;
Pictogram level search means for searching for the evaluation level for the extracted evaluation pictogram using the pictogram dictionary means;
An evaluation level determining means for determining an evaluation level of the sentence file based on a sum of evaluation levels of all the evaluation pictograms;
A program that causes a computer to function.

Images