JP5890385B2 - Data processing apparatus and data processing method - Google Patents

Description

  The present invention relates to a data processing apparatus and a data processing method for quantifying an object based on emotion parameters.

Conventionally, a content distribution apparatus that distributes content to a user is known (for example, see Patent Document 1).
In the device described in Patent Document 1, search conditions such as an author name, a title name, and a category are input at a user terminal and transmitted to a content distribution device. The content distribution apparatus has a database that associates the author name, title name, category, and the like of the content, extracts content that matches the input author name, title, category, and transmits it to the user's terminal for display.

JP 2007-65841 A

By the way, in the content distribution apparatus like the above-mentioned patent document 1, although the content can be searched by the author name, title name, category, etc., the target content cannot be searched unless the author name, title name, category is known. In other words, for example, when it is desired to examine content that is similar to an emotion that a person feels with respect to a predetermined content, such as when the atmosphere is similar to a certain content, there is a problem that a search cannot be performed with the above database.
Thus, there is a demand for an apparatus that can perform data processing around human emotions, such as when searching for content that has similar emotions that humans feel.

  An object of the present invention is to provide a data processing device and a data processing method capable of performing data processing around human emotion.

The data processing apparatus of the present invention uses a dictionary acquisition means for acquiring an emotion classification dictionary in which a plurality of words are classified in emotion units, and quantified data obtained by quantifying human emotions with respect to an object using the emotion classification dictionary. comprising a content quantification means for generating, wherein the dictionary acquisition means includes a dictionary generating means for generating said emotion classification dictionary based on text-based evaluation data for a plurality of objects, said dictionary generating unit, wherein From the evaluation data, the first element in the object, a plurality of the first phrases that have a high appearance frequency with respect to the main element and are within a predetermined range centered on a description position of the main element in the evaluation data, Is extracted by morphological analysis, and the first phrase that has a co-occurrence degree with respect to the main element equal to or greater than the predetermined value among the plurality of first phrases is defined as the second phrase. Extracted, characterized by classifying said second word for each emotion.

  In the present invention, quantification data for an object is generated using the emotion classification dictionary as described above. For this reason, by using quantified data centered on the emotions of such people, various data processing can be performed centering on emotions. For example, even if the author name or content name is unknown, it is the same as a certain content It is possible to perform processing such as search processing for searching for trendy content and display processing for displaying how a person feels as a review for a predetermined content.

1 is a block diagram showing a schematic configuration of a data processing system according to a first embodiment of the present invention. The block diagram which shows schematic structure of the server apparatus which is a data processor of 1st embodiment. The flowchart which shows the dictionary production | generation process of the emotion classification dictionary of 1st embodiment. The figure which shows an example of the review article on which evaluation data was published. The figure which shows an example of a co-occurrence network. The flowchart which shows the content quantification process of 1st embodiment. The flowchart which shows the content search process in 1st embodiment. The figure which compares the emotion value of the content searched by the search process, and query content. The flowchart which shows the content quantification data output process in 1st embodiment.

[First embodiment]
Hereinafter, a first embodiment according to the present invention will be described with reference to the drawings.
[overall structure]
FIG. 1 is a block diagram showing a schematic configuration of a data processing system according to the first embodiment.
As shown in FIG. 1, the data processing system 1 of this embodiment includes a user terminal 10 and a server device 20 that functions as the data processing device of the present invention. The user terminal 10 and the server device 20 are connected to a network. (For example, it is connected so that communication is possible via WAN (Wide Area Network), such as the internet.).
In this data processing system 1, the server device 20 displays content similar to the atmosphere of predetermined content (in this embodiment, content such as a book is exemplified as a target object) based on the search query received from the user terminal 10. Search and return the search result to the user terminal 10. In the data processing system 1, when the server device 20 receives a request for introduction of predetermined content from the user terminal 10, the server device 20 returns review data that quantifies emotions felt by a plurality of users for the content. And output (display) from the user terminal 10.
Hereinafter, a specific configuration and method for providing the above service will be described.

[User terminal configuration]
The user terminal 10 is a computer and includes a terminal communication unit 11, an input operation unit 12, a terminal storage unit 13, a terminal control unit 14, and a display 15, as shown in FIG.

The terminal communication unit 11 is connected to a network via, for example, a LAN, and communicates with other devices on the network.
The input operation unit 12 outputs an operation signal generated by a user operation to the terminal control unit 14. Examples of the input operation unit include a touch panel provided integrally with the display 15 and an input device such as a keyboard and a mouse.

The terminal storage unit 13 is configured by a data recording device such as a memory or a hard disk. The terminal storage unit 13 stores various programs for controlling the user terminal 10 and the like.
The terminal control unit 14 includes an arithmetic circuit such as a CPU (Central Processing Unit) and a storage circuit such as a RAM (Random Access Memory), and controls each unit of the user terminal 10. The terminal control unit 14 develops a program (software) such as a predetermined application from among the programs stored in the terminal storage unit 13 and the like, and performs various processes in cooperation with the program expanded in the RAM. Run. As a result, the terminal control unit 14 can communicate with the server device 20 via the network. For example, various services provided by the server device 20 can be used and various data can be viewed.
In addition, the terminal control unit 14 generates a search request or a content introduction request based on the search query and transmits it to the server device 20 by the user's operation of the input operation unit 12. These search queries and introduction requests include content specifying data for specifying content such as the title name of the content.
Furthermore, the terminal control unit 14 performs processing for displaying (outputting) various data transmitted from the server device 20 on the display 15.

[Configuration of server device]
FIG. 2 is a block diagram showing the server device 20 of the present embodiment.
The server device 20 according to the present embodiment is a computer, and includes a communication unit 21, a storage unit 22 (storage unit), a control unit 23, and the like.
The communication unit 21 is connected to a network via, for example, a LAN and communicates with other devices on the network.

The storage unit 22 is a data recording device constituted by, for example, a memory, a hard disk, etc., and constitutes a data storage means in the present invention.
The storage unit 22 stores various programs and various data for controlling the search device. The storage unit 22 stores an emotion classification dictionary and quantified data obtained by quantifying content based on the emotion classification dictionary.
In the present embodiment, an example is shown in which the storage unit 22 of the server device 20 functions as a data storage unit. For example, the data storage unit is provided in another device on the network, and the quantified data is stored. May be. The same applies to the emotion classification dictionary, and the emotion classification dictionary may be recorded in a predetermined device on the network.
In addition, a character dictionary is recorded in the storage unit 22. This character dictionary is a dictionary in which content and characters appearing in the content are associated with each other.

[Emotion classification dictionary stored in memory]
The emotion classification dictionary is a dictionary generated based on evaluation data for content. Table 1 shows an example of the emotion classification dictionary of this embodiment.

As shown in Table 1, the emotion classification dictionary is a dictionary in which emotion topics indicating human emotions and words and phrases related to the emotion topics are associated with each other. Here, the phrase is preferably an emotion expression phrase representing human emotion, such as an adjective or adjective verb, as well as a noun or verb representing emotion.
A detailed method for generating the emotion classification dictionary as described above will be described later.

[Quantification data]
The quantification data is data obtained by quantifying emotions felt by humans for each content using the emotion classification dictionary, and the quantification data for each content is recorded in a quantification database as shown in Table 2. .

In Table 2, the content ID is data for identifying the content and specifying the content. In the present embodiment, the content ID is exemplified, but, in addition, for example, the title name of the content may be used as data for specifying the content.
The emotion value is set for each of the emotion topics, and is a value obtained by quantifying the emotion of each emotion topic. A detailed method for generating quantified data will be described later.

  The control unit 23 includes an arithmetic circuit such as a CPU and a storage circuit such as a RAM. The control unit 23 expands a program (software) stored in the storage unit 22 or the like into the RAM, and cooperates with the program expanded in the RAM. Various processes are executed. And the control part 23 performs the said various processes, and as shown in FIG. 2, the evaluation data acquisition means 231, the data analysis means 232, the dictionary acquisition means 233, the quantification means 234, the request | requirement acquisition means 235, search It functions as the means 236, the quantified data output means 237, and the like.

The evaluation data acquisition unit 231 acquires evaluation data. Specifically, the evaluation data acquisition unit 231 is a device that provides another device, for example, an SNS (Social Networking Service), a device that publishes a content review site or blog site, Twitter (registered trademark). It is acquired from a device that publishes user's speech data on the network in real time. That is, it acquires text-based data such as personal blogs and review articles that can be browsed on the network, and user comments (tweets).
The data analysis unit 232 performs morphological analysis on the acquired evaluation data, and extracts a phrase (first phrase) included in the evaluation data.

The dictionary acquisition unit 233 generates an emotion classification dictionary as shown in Table 1 using the words (first words) extracted from the evaluation data.
Specifically, the dictionary acquisition means 233 includes co-occurrence determination means 233A and classification means 233B that function as dictionary generation means of the present invention. The co-occurrence determination unit 233A determines the co-occurrence degree between the extracted words (first words) and extracts the co-occurrence word (second word) based on the co-occurrence degrees. The classifying unit 233B classifies (clusters) the second phrase for each emotion topic.

The quantification unit 234 functions as the content quantification unit of the present invention. The quantification unit 234 acquires (reads) the emotion classification dictionary from the storage unit 22, quantifies emotions and atmospheres that people feel about the content, and generates quantified data.
The request acquisition unit 235 functions as a query acquisition unit and acquires various requests transmitted from the user terminal 10. In the present embodiment, as an example, a search request including a search query, an output request for requesting output of quantified data for a predetermined content, and the like are illustrated. As the search query, data specifying a predetermined content, for example, a content name is specified.

When the search unit 236 receives a search request, the search unit 236 searches for content whose tendency of the quantified data is similar to the content specified in the search query.
When receiving the output request, the quantified data output unit 237 reads the quantified data of the designated content and transmits it so that it can be displayed on the user terminal 10.
Detailed processing of each functional configuration will be described later.

[Data processing method]
Next, a data processing method in the data processing system 1 as described above will be described with reference to the drawings.

(Dictionary generation process)
FIG. 3 is a flowchart showing the emotion classification dictionary generation processing in the server device 20.
The server device 20 first generates an emotion classification dictionary in order to generate quantified data that quantifies human emotions regarding the content.

For this, the evaluation data acquisition means 231 of the server device 20 acquires evaluation data for a plurality of contents from the network (step S11).
In this step S11, it is preferable to obtain evaluation data for a plurality of contents having different tendencies. For example, when movies are targeted as contents, a plurality of contents having different release dates, director names, and categories (for example, romance movies, historical movies, SF movies, action movies, etc.) are targeted. These contents may be appropriately set by the administrator of the server device 20, for example, referring to a content introduction site, etc., each time data related to new contents is published, the contents name is obtained and evaluation data It may be an acquisition target.

Further, the evaluation data acquisition unit 231 acquires the evaluation data by using the content name of the content to be acquired as a query, and Web data such as a review article, a blog, and a user's remarks in which the evaluation data is described from the Internet. Search for. FIG. 4 is a diagram illustrating an example of a review article on which evaluation data is posted.
The evaluation data acquisition unit 231 analyzes the Web data 40 such as a review article searched as shown in FIG. 4 and acquires evaluation data 41 described as text data. As a method for obtaining the evaluation data 41 from the Web data 40, a known analysis method can be used. For example, a markup language such as html (HyperText Markup Language) is analyzed to extract a text description portion.
The evaluation data acquisition unit 231 preferably acquires a plurality of evaluation data for each content.

Next, the data analysis means 232 performs a morphological analysis on the evaluation data acquired in step S11, and extracts a phrase (step S12).
Thereafter, the co-occurrence determination unit 233A of the dictionary acquisition unit 233 reads the character dictionary from the storage unit 22, and the appearance frequency is high for the characters appearing in the content among the first words extracted in step S12. An emotion expression phrase representing emotion is extracted (step S13). Such a character becomes a main element in the present invention, and extracting an emotion expression word / phrase having a high appearance frequency with respect to the main element is a word indicating a strong feeling for the character of the creator who generated the evaluation data. .
Specifically, in step S <b> 13, the co-occurrence determination unit 233 </ b> A reads a character corresponding to the content from the character dictionary stored in the storage unit 22. Then, a description position where the character is described in the evaluation data is specified, and an emotion expression word / phrase appearing at a position close to the description position is extracted. For example, a sentence in which a character is described is specified, and emotion expression phrases are extracted from the specified sentence and the sentences before and after the specified sentence. Examples of emotion expression phrases include adjectives and adjective verbs, and nouns and verbs indicating human emotions. Particularly, adjectives and adjective verbs that are easy to specify are preferable.

Then, the co-occurrence determining unit 233A calculates the co-occurrence degree between the extracted characters and the emotion expression words / phrases for the characters (step S14).
In step S14, the co-occurrence determining unit 233A uses, for example, the number of times the words co-occurred (number of appearances) for the characters as the degree of co-occurrence of the words for the characters. As the co-occurrence degree, the Jaccard coefficient of the characters and phrases may be used as the co-occurrence degree.
In step S14, the co-occurrence determining unit 233A may construct a co-occurrence network 50 as shown in FIG. FIG. 5 is a diagram illustrating an example of the co-occurrence network 50.
In FIG. 5, 51 is a character, 52 is an emotion expression word (first word) extracted for the character, and indicates a co-occurrence relationship by lines connecting the words. Here, in FIG. 5, the thicker the line width, the higher the co-occurrence degree.
In the present embodiment, it is possible to calculate the co-occurrence degree with higher accuracy by determining the co-occurrence degree as described above based on a plurality of evaluation data.

Thereafter, the co-occurrence determining unit 233A extracts an emotion expression phrase having a co-occurrence degree equal to or greater than a predetermined value from the emotion expression phrases (first phrase) for the character as a second phrase (step S15).
When a co-occurrence network 50 as shown in FIG. 5 is constructed, a phrase whose line width is equal to or greater than a predetermined value is extracted as a second phrase.

Next, the classification unit 233B of the dictionary acquisition unit 233 classifies (clusters) the second word / phrase extracted in step S15 for each emotion topic (cluster), and creates an emotion classification dictionary that associates the word / phrase corresponding to each emotion topic. Create (step S16).
Here, the classification means 233B uses LDA (Latent Dirichlet Allocation) as a classification method of the extracted second word / phrase. As a result, an optimal number of emotion topics are calculated based on the extracted second phrase, and the similarity between each emotion topic and the second phrase (probability that the second phrase is related to the emotion topic) is calculated. . Therefore, an emotion classification dictionary as shown in Table 1 can be created by associating each emotion topic with a second phrase having a predetermined similarity or higher.
In the present embodiment, an example of automatically generating emotion topics and the number thereof by LDA has been shown. However, the present invention is not limited to this, and the number of emotion topics or emotion topics to be set may be set in advance. .

The implementation timing of the dictionary generation process described above may be, for example, a timing designated by the administrator of the server device 20, for example, once a month, etc. The classification dictionary may be updated as needed.
Alternatively, predetermined Web data (such as a content review site) on the Internet may be monitored, and an emotion classification dictionary may be created and updated each time data relating to new content is released.

(Content quantification process)
Next, content quantification processing will be described with reference to the drawings.
FIG. 6 is a flowchart showing content quantification processing in the server device 20.
The server device 20 uses the emotion classification dictionary generated as described above to generate quantified data that quantifies human emotions with respect to the content.

For this purpose, the quantification means 234 of the server device 20 identifies the content that is the target for generating the quantification data (step S21).
In step S21, the content may be specified by monitoring predetermined Web data (for example, a content review site) on the Internet, for example, and acquiring the content name every time data related to new content is released. The update status of the Web data may be periodically acquired, and the content name may be acquired every time data related to new content is released through the update. For example, the server administrator may specify the content by inputting data specifying the content such as a content name at a timing specified by the administrator of the server device 20.

Thereafter, the evaluation data acquisition unit 231 acquires evaluation data for the content specified in step S21 from the network (step S22).
In this step S22, the evaluation data acquisition means 231 acquires the evaluation data by using the content name of the identified content as a query and Web data (for example, review articles, blogs, user's Search).
Then, the evaluation data acquisition means 231 analyzes these review articles, blogs, user comments, etc., and acquires evaluation data of text data. As in step S12, the evaluation data acquisition unit 231 preferably acquires a plurality of evaluation data for the content.

Next, the data analysis means 232 performs morphological analysis on the evaluation data acquired in step S22, and extracts words included in the evaluation data (step S23).
Thereafter, the quantification means 234 reads the emotion classification dictionary generated by the dictionary generation process and stored in the storage unit 22, and acquires the emotion value for each emotion topic of the content (step S24).
Specifically, the quantifying means 234 acquires the number of phrases extracted in step S23 included in each emotion topic of the emotion classification dictionary as an emotion value. That is, the total number of words included in the emotion topic is used as the emotion value. For example, when the phrase extracted in step S23 is “pounding”, “pounding”, “power”, and using the emotion classification dictionary of Table 1 described above, the quantifying means 234 determines that the emotion is “awakening (pounding)”. Since the words “pounding” and “powerful” are included for the topic, the emotion value for the emotional topic “awakening” is set to “3”.
When the emotion value for each emotion topic is set in step S24, the quantification means 234 obtains the emotion value for each emotion topic and content specifying data (for example, content ID or content name) for specifying the content. The associated quantification data is generated and stored in the storage unit 22 (step S25).

(Content search process)
Next, content search processing will be described with reference to the drawings as an example of a service using quantified data as described above.
FIG. 7 is a flowchart showing content search processing in the present embodiment.
In the user terminal 10, when the input operation unit 12 is operated by the user and a search query is input, the terminal control unit 14 of the user terminal 10 generates a search request including the search query (step S31). Here, as the search query, content specifying data (for example, content name, content ID, etc.) specifying the content is specified. Further, the user terminal 10 transmits a user ID for identifying the user terminal 10 together with the search request (step S32).

When receiving the search request transmitted from the user terminal 10 by the request acquisition unit 235 (step S41), the server device 20 causes the search unit 236 to perform content search processing.
In the content search process, first, the search unit 236 determines whether or not there is quantification data corresponding to content specified by the search query included in the search request (hereinafter sometimes referred to as query content). (Step S42).
When it is determined as “No” in step S42, the search unit 236 outputs a non-corresponding output instruction to the user terminal 10 to output that the query content is not found on the display 15 (step S43).

When it is determined as “Yes” in step S42, the search unit 236 acquires quantification data of the query content (step S44).
Then, the search means 236 searches the quantification database for quantification data whose tendency is similar to the quantification data acquired in step S44, and acquires the content name of the content and the quantification data of the content (step S45). .
Thereafter, the search unit 236 returns the content name and quantification data acquired in step S45 to the user terminal 10 (step S46).

The terminal control unit 14 of the user terminal 10 determines whether a non-corresponding output instruction has been received from the server device 20 or a search result has been received (step S33). If a non-corresponding output instruction is received in step S33, the display 15 displays that the corresponding content is not found (step S34).
If the search result is received in step S33, the searched content name and the quantified data of the content are displayed on the display 15 as the search result (step S35).
FIG. 8 is a diagram illustrating an example in which the quantified data in the query content and the searched content is compared. The solid line in FIG. 8 indicates the searched content, and the broken line indicates the query content. As shown in FIG. 8, content similar in tendency to the query content and each emotion value is searched by the above search processing.

(Content quantification data output processing)
Next, content quantification data output processing will be described with reference to the drawings as another example of the service using the quantification data as described above.
FIG. 9 is a flowchart showing content quantification data output processing in the present embodiment.
In the user terminal 10, when the user operates the input operation unit 12 and an output request for quantified data for a predetermined content is input, the terminal control unit 14 transmits the output request and the user ID to the server device 20. (Step S51).

When receiving the search request transmitted from the user terminal 10 by the request acquisition unit 235 (step S61), the server device 20 causes the quantified data output unit 237 to execute content quantification data output processing.
In the content quantification data output process, first, the quantification data output means 237 determines whether or not there is quantification data corresponding to the content specified in the output request (step S62).
If it is determined as “No” in step S62, a non-corresponding output instruction is output to the user terminal 10 as in step S43.
When it is determined as “Yes” in step S62, the quantified data output unit 237 acquires the quantified data of the designated content (step S63) and returns it to the user terminal 10 (step S64).

The terminal control unit 14 of the user terminal 10 determines whether a non-corresponding output instruction is received from the server device 20 or quantified data is received (step S52). If a non-corresponding output instruction is received in step S52, the display 15 displays that the corresponding content is not found, as in step S34.
In step S52, when quantification data for the designated content is received, the quantification data is displayed on the display 15 (step S53).

[Operational effects of the first embodiment]
The quantification unit 234 of the server device 20 according to the present embodiment acquires an emotion classification dictionary in which a plurality of words are classified by emotion topic unit from the storage unit 22 and uses this emotion classification dictionary to express a person's feelings about the content. Generate quantified data.
In other words, since the emotion classification dictionary is associated with emotion topics for words indicating human emotions, using such a dictionary, it is possible to analyze and express what impression a person has with respect to content. Can be quantified. By using such quantified data, for example, various types of information processing based on human emotions such as the above-described search processing and content quantified data output processing can be performed.

  In the present embodiment, the dictionary acquisition unit 233 generates an emotion classification dictionary based on evaluation data for content of a plurality of users (evaluators) disclosed on the network (Internet). That is, the dictionary acquisition unit 233 generates an emotion classification dictionary based on emotions felt by individual users with respect to the content. In this way, by generating an emotion classification dictionary around human emotions, for example, it is possible to generate quantified data that matches human emotions compared to mechanically classifying words and phrases, and various processes such as search processing The processing accuracy can be improved.

In the present embodiment, the data analysis unit 232 extracts a plurality of first words / phrases from the evaluation data by morphological analysis, and the co-occurrence determination unit 233A of the dictionary acquisition unit 233 determines that the co-occurrence degree of the extracted first words / phrases has a predetermined value. The first word / phrase is extracted as the second word / phrase.
In the evaluation data, a phrase with a high degree of co-occurrence is a phrase in which the person who created the evaluation data has a strong feeling for the content, and is often associated with the same feeling. Therefore, by extracting words having a high degree of co-occurrence, it is possible to generate a highly accurate emotion classification dictionary in which appropriate phrases are associated with emotion topics.

  In the present embodiment, the co-occurrence determining unit 233A extracts characters that are main elements in the evaluation data and words / phrases for the characters. Such a phrase is a phrase that indicates the emotion that the evaluator who generated the evaluation data strongly holds for the character, and is likely to be a phrase that strongly reflects the image that the evaluator has on the content . Therefore, by creating an emotion classification dictionary based on these phrases, it is possible to generate an accurate emotion classification dictionary in which appropriate phrases are associated with emotion topics.

  At this time, the co-occurrence determining unit 233A identifies a sentence in which the characters in the evaluation data are described, and extracts an emotion expression phrase from the identified sentence and the sentences before and after the sentence. In other words, phrases described within a predetermined range centering on the characters are extracted. Thereby, the phrase which shows the emotion with respect to a character can be extracted more accurately.

In the present embodiment, emotion expression phrases such as adjectives and adjective verbs are extracted from the evaluation data.
Thereby, an optimal phrase can be associated with each emotion topic in the emotion classification dictionary.

In the present embodiment, the classifying unit 233B classifies the extracted phrases for each emotion topic using LDA. By using the LDA, it is possible to automatically classify the optimal emotion topic by the optimal number of clusters in the second word / phrase extracted by the co-occurrence determination unit 233A.
As a result, the process in the dictionary generation process can be simplified and speeded up, and the accuracy of the emotion classification dictionary can be improved.

In the present embodiment, the quantification unit 234 quantifies the content based on the evaluation data for the content acquired by the evaluation data acquisition unit 231 and the emotion classification dictionary.
In other words, based on the evaluation data for the content of a plurality of users (evaluators), the quantification data of the content is generated based on the emotion felt by each user for the content. For this reason, for example, rather than generating quantification data based on the content (for example, synopsis), appropriate quantification data based on emotions can be generated based on human emotion evaluation.

  In the present embodiment, the quantification data generated by the quantification unit 234 is accumulated in the storage unit 22. Therefore, various processes can be easily performed by reading out the quantification data.

In the present embodiment, when the request acquisition unit 235 receives a search request including a search query in which content (query content) is specified, the search unit 236 has quantification data similar to the quantification data of the query content. The content is retrieved from the quantification database and returned to the user terminal 10.
Conventionally, when searching for content with the same atmosphere (content with the same impression and emotion received by people) for a certain query content, the content of the same author as that content is searched and displayed, or the content is purchased There are known services for searching and displaying what kind of content other people have purchased. However, in such a search service, the searched content does not always have the same atmosphere as the query content. On the other hand, in this embodiment, since quantified data obtained by quantifying human emotions for each content based on the emotion classification dictionary is used, content similar to the query content (the same emotion as the query content is held). Can be suitably searched.

  In this embodiment, when the output request for the content is acquired by the request acquisition unit 235, the quantified data output unit 237 acquires the quantified data for the content and returns it to the user terminal 10. Thereby, emotion quantification data for the content is displayed on the user terminal 10. In this way, by displaying emotions felt by a plurality of evaluators with respect to the content using, for example, a radar chart as shown in FIG. 8, the user can easily understand the evaluation of the content.

[Second Embodiment]
In the first embodiment described above, the quantification unit 234 generates quantification data based on evaluation data from a plurality of evaluators for one content. In this case, depending on the person, how to feel the content differs, for example, a group of evaluators who feel “scary” for one content (emotion group) and an emotion group of evaluators who feel “interesting” are mixed. there's a possibility that. As described above, when a plurality of emotion groups coexist, if the emotion topic is quantified, the feature of the content may be difficult to see or may be quantified data that shows a feature different from the feature of each emotion group.

  On the other hand, in the second embodiment, in order to solve the above-described problem, the quantification unit 234 quantifies data that associates the co-occurrence relationship between emotion topics with emotion values for each content. Is generated. Specifically, the quantification unit 234 obtains a co-occurrence relationship between emotion topics using a technique such as cross tabulation, correlation analysis, or multidimensional analysis.

As described above, even when there are a plurality of emotion groups for the content by using the quantification data in which the co-occurrence relationship between the emotion topics is associated in addition to the emotion value of each emotion topic as the quantification data ( Even if the feeling of content differs among people), it becomes easier to see the characteristics of the content and also to understand the characteristics of each emotion group.
For example, if the emotional topic of “Scared” and the emotional topic of “Interesting” are co-occurring with respect to the content, by looking at the quantification data of the content, You can see that there are people who feel “interesting”. In this case, it can be distinguished from the case where the user feels both “scary” and “interesting” at the same time, and can provide quantified data that makes the characteristics of each content easier to understand.

[Modification]
In addition, this invention is not limited to embodiment mentioned above, In the range which can achieve the objective of this invention, the deformation | transformation shown below is also included.
[Modification 1]
In the above-described embodiment, the server device 20 returns the non-corresponding output instruction to the user terminal 10 when it is determined “No” in step S42. However, the present invention is not limited to this. For example, the content quantification process described above may be performed on the content specified by the search query. In this case, the content specified in step S21 is the content specified by the search query, evaluation data is acquired from the network, and the content is quantified based on the evaluation data.

[Modification 2]
In the above embodiment, an example is shown in which content quantification data is displayed based on an output request from the user terminal 10, but the present invention is not limited to this. For example, the present invention can be applied to various services using quantified data, such as obtaining and posting quantified data for the content from the server device 20 at an introduction site or the like that introduces content.

[Modification 3]
In the said embodiment, although content, such as a book and a movie, was illustrated as a target object of this invention, it is not limited to this. The target object may be any target as long as it can be evaluated by the user (evaluator). For example, when applied to restaurants such as restaurants, emotion classification dictionaries and quantified data of each store can be generated based on gourmet report articles such as store atmosphere and taste trends. In this case, when a predetermined store name is input as a search query, a store having a similar atmosphere can be searched. In addition, by using these quantified data at a review site of a restaurant, etc., it is possible to expand the use, for example, by building a site that can intuitively understand the tendency of each store.

[Modification 4]
In the above embodiment, the dictionary acquisition unit 233 extracts the evaluation data from the Web data on the network acquired by the evaluation data acquisition unit 231 and generates the emotion classification dictionary based on the evaluation data. However, the present invention is not limited to this. . For example, the emotion classification dictionary may be generated based on the input evaluation data by inputting the evaluation data aggregated by a questionnaire or the like to the server device 20. The same applies to the quantifying means 234, and the acquisition source of the evaluation data is not limited to the evaluation data published on the network.

[Modification 5]
The co-occurrence determination unit 233A of the dictionary acquisition unit 233 extracts the second word / phrase based on the number of co-occurrence as the degree of co-occurrence. For example, as described above, the second word / phrase is extracted based on the Jaccard coefficient or the like. May be.
In addition, although the accuracy is lowered, the emotion classification dictionary may be extracted as the second phrase without being limited to the co-occurrence degree.

[Modification 6]
In the above embodiment, the dictionary acquisition unit 233 has shown an example of extracting words / phrases having a high appearance frequency with respect to the characters, but the present invention is not limited to this. For example, in order to extract a person's feelings and the like for the entire content work, the content name, content creator, and the like may be used as main elements, and words for these main elements may be extracted.
Moreover, you may extract not only a main element but an emotion expression phrase. For example, when analyzing evaluation data of a restaurant or a product user such as a store or a product as an object, an emotion expression phrase may be extracted with the evaluator itself (for example, my first person subject) as a main element.

[Modification 7]
The classification unit 233B classifies the second word / phrase extracted by the LDA, but is not limited thereto. For example, a classification table in which emotion topics for words are associated in advance may be generated, and classification based on the classification table may be performed.

[Modification 8]
In the above embodiment, an example in which the evaluation data acquisition unit 231, the data analysis unit 232, the dictionary acquisition unit 233, and the quantification unit 234 are provided in the server device 20 is shown, but the present invention is not limited to this. For example, the terminal control unit 14 of the user terminal 10 reads and executes a program stored in the terminal storage unit 13, so that the evaluation data acquisition unit 231, the data analysis unit 232, the dictionary acquisition unit 233, and the quantification unit 234 are used. It may be configured to function. In this case, the quantification data of the application installed in the user terminal 10 can be displayed. It is also possible to determine user preference data obtained by collecting quantification data of installed applications. By transmitting such preference data to the application providing device and the advertisement distribution device, it is possible to distribute advertisements and application introductions that are useful to the user.

[Modification 9]
In the above embodiment, the quantification unit 234 acquires the number of words classified into each emotion topic itself as an emotion value, but is not limited thereto, and uses other emotion value setting methods based on the classification result. May be.
For example, the ratio of the number of phrases classified with respect to each emotion topic with respect to the total number of phrases extracted in step S23 (occupation ratio of each emotion topic with respect to the whole) and the distribution may be used as the emotion value.
In addition, for example, by detecting the number of phrases classified for each emotion topic every predetermined period, the transition (change rate) of the number of phrases classified for each emotion topic may be calculated and used as an emotion value.

[Modification 10]
Although the dictionary acquisition unit 233 includes the co-occurrence determination unit 233A and the classification unit 233B and functions as the dictionary acquisition unit, the example is not limited thereto. For example, the dictionary acquisition unit 233 may acquire the emotion classification dictionary from another device on the network.

  In addition, the specific structure and procedure for carrying out the present invention can be appropriately changed to other structures and the like within a range in which the object of the present invention can be achieved.

  DESCRIPTION OF SYMBOLS 1 ... Data processing system, 10 ... User terminal, 20 ... Server apparatus (data processing apparatus), 21 ... Communication part, 22 ... Memory | storage part (memory | storage means), 23 ... Control part, 41 ... Evaluation data, 50 ... Co-occurrence network 231 ... Evaluation data acquisition means, 232 ... Data analysis means, 233 ... Dictionary acquisition means 233, A ... Co-occurrence determination means, 233B ... Classification means, 234 ... Quantification means, 235 ... Request acquisition means, 236 ... Search means, 237: Quantified data output means.

Claims (10)

A dictionary acquisition means for acquiring an emotion classification dictionary in which a plurality of words are classified by emotion unit;
Using the emotion classification dictionary, content quantification means for generating quantification data quantifying human emotions on the object,
The dictionary acquisition means includes dictionary generation means for generating the emotion classification dictionary based on text-based evaluation data for a plurality of objects ,
The dictionary generation means includes, from the evaluation data, a plurality of elements that are within a predetermined range centered on a main element in the target object, a frequency of appearance of the main element, and a description position of the main element in the evaluation data. And the first word / phrase by morphological analysis, and the first word / phrase having a co-occurrence degree with respect to the main element equal to or greater than the predetermined value is extracted as the second word / phrase among the plurality of first words / phrases. A data processing device that classifies the second phrase for each emotion . The data processing apparatus according to claim 1 ,
The data processing apparatus, wherein the first word / phrase extracted by the morphological analysis is an emotion expression word / phrase indicating a human emotion. In the data processing device according to claim 1 or 2 ,
The dictionary generation means classifies the extracted second word / phrase for each emotion using a potential Dirichlet distribution method. The data processing device according to any one of claims 1 to 3 ,
Comprising evaluation data acquisition means for acquiring text-based evaluation data for the object;
The content quantifying means generates the quantified data based on evaluation data for the object and the emotion classification dictionary. The data processing apparatus according to claim 4 , wherein
The content quantifying means classifies a phrase extracted by morphological analysis of the evaluation data for the target object for each emotion using the emotion classification dictionary, and associates an emotion value based on a classification result with the target object. A data processing device characterized by generating the quantification data. The data processing apparatus according to claim 5 , wherein
The content quantifying means generates quantified data in which a co-occurrence relationship between the emotions is associated with the emotion value. The data processing apparatus according to any one of claims 1 to 6 ,
A data processing apparatus comprising data storage means for storing the quantification data generated by the content quantification means. The data processing apparatus according to claim 7 , wherein
Query acquisition means for acquiring a predetermined object as a search query;
Search means for searching quantification data similar to the quantification data for the object specified as the search query from the data storage means, and returning an object corresponding to the searched quantification data as a search result;
A data processing apparatus comprising: In the data processing device according to claim 7 or 8 ,
Request acquisition means for acquiring an output request for the quantification data for a predetermined object;
Quantified data output means for acquiring and outputting the quantified data for the object specified as the output request from the data storage means;
A data processing apparatus comprising: A data processing method for generating quantified data by quantifying emotions felt by a person with respect to an object,
The computer
Generating an emotion classification dictionary in which a plurality of phrases are classified by emotion unit based on text-based evaluation data for a plurality of objects, and storing it in a storage means ;
Obtaining the emotion classification dictionary from the storage means ;
Using the emotion classification dictionary to generate the quantification data for the object ,
In the step of generating the emotion classification dictionary and storing it in the storage means, from the evaluation data, the main element in the object, the frequency of appearance with respect to the main element, and the description position of the main element in the evaluation data A plurality of the first words and phrases within a predetermined range centered on the morpheme analysis, and among the plurality of first words and phrases, the co-occurrence degree with respect to the main element is equal to or greater than the predetermined value. A data processing method characterized by extracting a phrase as the second phrase and classifying the second phrase for each emotion .

Images