JP5928091B2 - Tag group classification method, apparatus, and data mashup method, apparatus - Google Patents

Description

  The present invention relates to data processing, and more specifically, to a tag group classification method and apparatus, and a data mashup method and apparatus.

  Currently, various data format specifications for describing data, such as XML (eXtensible Markup Language), JSON (JavaScript (registered trademark) Object Notation), CSV ( Comma Separated Values). Each of the various data format specifications defines a tag for describing the meaning of the data content. For example, for a list type data, for example, a news list including some news, tags of one group for describing the news contents, that is, title (title), pubdate (announcement time), author (author), etc. Can be defined. Further, for example, for a schedule table including a few schedules, tags of one group for describing schedule contents, that is, starttime (end time), endtime (endtime), attendees (participant), and location ( Place) etc. can be defined. Therefore, the data contents can be conveniently announced or accessed using the tag.

  However, data contents having the same or similar meaning may be described by using different tags with different data format specifications. For example, for data content “data creator”, different tags such as “author”, “writer” or “creator” may be employed in different data format specifications. Therefore, by recognizing the data contents describing the same or similar meaning with different tags and describing the same or similar data contents with a uniform tag, mashup of the data contents having the same or similar meaning There is a hope to complete.

  In the conventional technique, it is determined whether or not the plurality of data contents are the same or similar by directly comparing the plurality of data contents themselves. Since the amount of data of the data content itself is large, when a plurality of data contents themselves are directly compared, the amount of calculation becomes large, and the accuracy of judgment often deteriorates.

  In addition, in the conventional technique, there is also a technique for determining whether or not the data contents described by two tags are the same or similar by comparing whether or not the two tags are the same or similar. However, in actual applications, there are various different data format specifications, and the adopted tags are different. If only tags are compared, it will be difficult to comprehensively consider various features of various tags, and the accuracy of judgment will deteriorate.

  As described above, for example, for a news list including some news, one group tag (hereinafter referred to as “tag group”) for describing the content of one news, that is, title (Title), pubdate (announcement time), author (author), etc. can be defined. As a result, it has been found that one data content is generally defined by a tag group including a few tags for describing the data content. Therefore, in order to determine whether or not there is the same or similar meaning among a plurality of data contents, it is comprehensively determined whether or not a plurality of tag groups for describing a plurality of data contents are the same or similar. It is necessary to judge. If only tags are compared, it is difficult to determine whether or not the data contents described by a tag group including some tags have the same or similar meaning.

  In view of the above problems, the applicant considered that data contents having the same or similar meaning should be recognized by comparing whether or not a plurality of tag groups are the same or similar. The basic idea of the present invention is that, in order to compare whether or not a plurality of tag groups are the same or similar, first, the same or similar tag groups are classified into the same class and then newly discovered. The tag group is compared with the class of the segmented tag group. Since all tag groups in the same class are the same or similar, the tag group class comprehensively considers the various features of the various tag groups. Therefore, by comparing the tag group and the class of the tag group, it is possible to more accurately determine the same or similar between the tag groups.

  An object of the present invention is to provide a tag group classification method and apparatus, and a data mashup method and apparatus.

  According to one embodiment of the present invention, a method is provided for a computer to classify a tag group that includes at least one tag and corresponding data defined by the at least one tag. The computer shows how many tags belonging to a certain group appear in which synonymous tag from the synonymous tag set group to which the synonymous tag belongs and the tag group group to which the tag defining the data of one data list belongs. An element group is generated, a feature vector corresponding to each tag group is generated from the element group, and the tag group is classified into classes according to the similarity of the feature vectors of each tag group.

  According to at least one embodiment of the present invention, by comparing the similarity between a tag group feature vector and a tag group class core feature vector, more accurately and more efficiently between tag groups. The same or similar can be determined, and the same or similar data can be mashed up more accurately and more efficiently.

4 is a flowchart illustrating a method for classifying a tag group according to an embodiment of the present invention. 4 is a flowchart illustrating a specific procedure of a classification step of a method for performing classification on a tag group according to an embodiment of the present invention. FIG. 3 is a block diagram illustrating an apparatus for classifying tag groups according to an embodiment of the present invention. 6 is a flowchart illustrating a method for mashing data based on tag groups according to an embodiment of the present invention. FIG. 3 is a block diagram illustrating an apparatus for mashing up data based on tag groups according to an embodiment of the present invention. And FIG. 7 is a block diagram illustrating an exemplary structure of a computer that implements an embodiment of the present invention.

  The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used in this application, the singular forms “a” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. Also, the word “comprising” as used herein means the presence of the indicated feature, whole, step, operation, unit and / or component, but one or more other features. Does not exclude the presence or addition of whole, steps, operations, units and / or components, and / or combinations thereof.

  Embodiments of the present invention will be described below with reference to the drawings. It should be noted here that, for the sake of clarity, notations and descriptions of parts and processes known to those skilled in the art that are not related to the present invention have been omitted from the drawings and description. Each block in the flowchart and / or block diagram and a combination of each block in the flowchart and / or block diagram can be realized by instructions of a computer program. The instructions of these programs can be provided to the processor of a general purpose computer, special purpose computer or other programmable data processing device. This provides a machine that causes these instructions executed by a computer or other programmable data processing device to generate a device that implements the functions / operations defined by the blocks in the flowcharts and / or block diagrams. .

  The instructions of these computer programs may be stored on a computer readable medium that can be controlled to operate a computer or other programmable data processing device in a particular fashion. Thus, a product is provided that includes instructions means for implementing the functions / operations defined by the blocks in the flowcharts and / or block diagrams by means of instructions stored on a computer readable medium.

  Computer program instructions are loaded onto a computer or other programmable data processing device and a series of operational steps are performed on the computer or other programmable data processing device to generate a computer implementation. Thus, instructions executed on a computer or other programmable data processing device provide a process for implementing the functions / operations defined by the blocks in the flowcharts and / or block diagrams.

  The flowchart and block diagrams in the Figures illustrate the system architecture, functionality, and operation that can be implemented by systems, methods and computer program products according to various embodiments of the present invention. In this regard, each block in the flowcharts and block diagrams may represent a module, program segment, or portion of code. The module, program segment, or code portion includes one or more executable instructions for implementing a predetermined logic function. Also, in some replacement implementations, the functions marked in the blocks may occur differently from the order marked in the drawing. For example, two connected blocks may actually be executed in parallel, or may be executed in the reverse order, depending on the function involved. Each block in the block diagram and / or flowchart and the combination of blocks in the block diagram and / or flowchart may be realized by a dedicated hardware system for executing a predetermined function or operation, Alternatively, it may be realized by a combination of dedicated hardware and computer instructions.

  Hereinafter, a method for classifying tag groups according to an embodiment of the present invention will be described with reference to FIG. FIG. 1 is a flowchart illustrating a method for classifying tag groups according to an embodiment of the present invention. This method is performed in the apparatus shown in FIG.

  As shown in FIG. 1, the method starts at step 100. Next, in step 102, the synonymous tag set specifying unit 300 of the apparatus shown in FIG. 3 specifies a synonymous tag set to which each tag in the tag group belongs among a plurality of synonymous tag sets.

  The synonym tag set (S) is a set composed of one group of tags having the same or similar meaning (that is, synonym). As an example, there may be some synonymous tag sets as follows.

S1: author, creator, writer
S2: pubdate (publication time), publishdate (announcement time)
S3: URL (Uniform Resource Location Specifier), link
S4: summary, description
S5: event, title, what
S6: starttime, when (what time)
S7: where, location
...
Sn: who, attendees (participants)
However, n is an integer of 1 or more.

  The above synonymous tag sets are merely examples, and other synonymous tag sets may exist as necessary. Based on actual application experience, it can be determined in advance which tags have the same or similar meaning. In addition, in the process of use, a newly discovered tag having the same or similar meaning can be added to the synonymous tag set to dynamically update the synonymous tag set. For example, the synonym tag set may be provided in the form of a synonym dictionary. One skilled in the art can appreciate that the synonymous tag set may be provided in other ways, such as a database.

  The tag group (T) is a set composed of one group of tags for defining corresponding data in one data list. As an example, there may be some tag groups as follows.

T1: title, author, pubdate, summary
T2: title, publication date, creator, creator, description, URL (Uniform Resource Locator)
T3: title, link, writer, description
T4: title, link, writer, description
T5: event, starttime, endtime, location, attendees
T6: title, starttime, duration, where, where, attendes
...
Tp: what, where, who, when
However, p is an integer of 1 or more.

  The above tag groups are merely examples, and other tag groups may exist in actual applications. For example, different data format specifications (eg, XML, JSON, CSV, etc.) may define different tag groups, or data presenters may customize different tag groups as needed.

  For one new tag group, the synonymous tag set to which each tag in the new tag group belongs can be specified based on the above synonymous tag set. For example, with respect to the tag group T1, the tag “title” in the tag group T1 belongs to the synonymous tag set S5 (that is, the tag group T1 belongs to the synonymous tag set S5 in accordance with the order of the tags in the tag group T1. The tag “author” in the tag group T1 belongs to the synonym tag set S1 (that is, the number of tags belonging to the synonym tag set S1 in the tag group T1 is 1). The tag “publication time” in the tag group T1 belongs to the synonym tag set S2 (that is, the number of tags belonging to the synonym tag set S2 in the tag group T1 is 1), and the tag “summary ( (Summary) ”belongs to the synonymous tag set S4 (that is, the number of tags belonging to the synonymous tag set S4 in the tag group T1 is 1) in order. Further, with respect to the tag group T1, the number of tags belonging to the synonym tag set S1 is one in the tag group T1 and the synonym tags in the tag group T1 according to the order from the synonym tag set S1 to the synonym tag set Sn. The number of tags belonging to the set S2 is 1, the number of tags belonging to the synonymous tag set S3 in the tag group T1, 0, the number of tags belonging to the synonymous tag set S4 in the tag group T1, and the same meaning in the tag group T1. The number of tags belonging to the tag set S5 is 1, the number of tags belonging to the synonym tag set S6 in the tag group T1, and the tags belonging to the set from the synonym tag set S7 to the synonym tag set Sn in the tag group T1. It can be specified in turn that the numbers of are each 0. According to the same method, each tag in each tag group from the tag group T2 to the tag group Tp can respectively identify which tag set from the synonymous tag set S1 to the synonymous tag set Sn. it can.

  The method then proceeds to step 104. In step 104, the feature vector generation unit 302 of the apparatus shown in FIG. 3 generates a feature vector corresponding to the tag group. Of the generated feature vector, each element corresponds to a different synonymous tag set in a plurality of synonymous tag sets, and the value of each element is the number of tags belonging to the synonymous tag set corresponding to the element in the tag group. .

  Based on the result specified in step 102, a feature vector corresponding to the tag group can be generated. For example, for the tag group T1, the feature vectors A: (S5: 1, S1: 1, S2: 1, S4: 1) corresponding to the tag group T1 according to the identification result in the order of each tag in the tag group T1. ) Can be generated. Of each element, the part before the colon indicates the synonymous tag set corresponding to the element, and the part after the colon indicates the number of tags belonging to the synonymous tag set corresponding to the element in the tag group 1. For example, for the first element “S5: 1” of the feature vector A, “S5” indicates that the first element corresponds to the synonymous tag set S5, and “1” indicates the synonymous tag set in the tag group T1. This indicates that the number of tags belonging to S5 is one. In addition, for the tag group T1, the feature vector A ′: (S1: 1, S2: 1, S3) corresponding to the tag group T1 is determined according to the specific result in the order from the synonymous tag set S1 to the synonymous tag set Sn. : 0, S4: 1, S5: 1, S6: 0, S7: 0, ..., Sn: 0). Since the meaning of each part of each element is the same as that in the feature vector A, description thereof is omitted here. According to a similar method, feature vectors corresponding to each of the tag groups from the tag group T1 to the tag group Tp can be generated.

  The method then proceeds to step 106. In step 106, the similarity calculation unit 304 of the apparatus shown in FIG. 3 calculates the similarity between the feature vector and the core feature vector of each class of at least one class. The value of each element of the core feature vector of the class is the sum of the values of the corresponding elements in the corresponding feature vector of each tag group already classified into the class.

  A class is a set composed of one group of tag groups that are the same or similar to each other. That is, each tag group belonging to the same class is the same or similar. For example, whether or not the tag groups are the same or similar can be determined based on the cosine distance between the tag groups. Hereinafter, a process of calculating a cosine distance between tag groups will be described.

It is assumed that the feature vector A corresponding to the tag group T1 is generated by the step 104 and the feature vector B corresponding to the tag group T2 is generated. The feature vector A is indicated by (S1: fa1, S2: fa2,..., Sn: fan), and can be described by an abbreviation of (fa1, fa2,..., Fan). Is represented by (S1: fb1, S2: fb2,..., Sn: fbn) and can be described by an abbreviation of (fb1, fb2,..., Fbn). Sn represents the synonymous tag set Sn corresponding to the nth element in the feature vector A or the feature vector B, and fan represents the synonymous tag set Sn corresponding to the nth element in the feature vector A in the tag group T1. The number of belonging tags is indicated, and fbn indicates the number of tags of the synonymous tag set Sn corresponding to the nth element in the feature vector B in the tag group T2. The cosine similarity between the feature vector A corresponding to the tag group T1 and the feature vector B corresponding to the tag group T2 can be calculated by the following equation (1). That is,
Similarity (A, B) = (Σfak × fbk) / sqrt [(Σfak × fak) × (Σfbk × fbk)] Equation (1)
However, 1 ≦ k ≦ n and n is an integer of 1 or more.

  For a class composed of one group of tag groups, for example, the corresponding elements of the corresponding feature vectors of each tag group in the class can be accumulated to obtain the corresponding core feature vector of the class. it can. For example, class C has tag groups from tag group T1 to tag group Tm (m is an integer greater than or equal to 1) already classified into class C, and from tag group T1 to tag group Tm. Assuming that the corresponding feature vectors of the tag groups are from the feature vector A1 to the feature vector Am, the corresponding core feature vector AC of class C is expressed by the following equation (2).

AC = (Σfaj1, Σfaj2, ..., Σfajn) Equation (2)
However, 1 ≦ j ≦ m, m is an integer of 1 or more.

  After calculating the corresponding core feature vector AC of class C using equation (2), the corresponding feature vector ANE of the new tag group TNE and the corresponding core feature vector of class C using equation (1) above Similarity with AC can be calculated. If there are a plurality of classes, the degree of similarity between the corresponding feature vector ANE of the new tag group TNE and the corresponding core feature vector of each of the plurality of classes is calculated.

  The method then proceeds to step 108. In step 108, the tag group classification unit 306 of the apparatus shown in FIG. 3 classifies the tag group into an approximate class among at least one class based on the calculated similarity.

  The magnitude of the value of the cosine similarity between the corresponding feature vector of the tag group calculated by the equation (1) and the corresponding core feature vector of the class indicates the similarity between the tag group and the class. As the cosine similarity value is larger, the tag group and the class are more similar. Therefore, it can be determined whether the tag group and the class are similar based on the calculated similarity, and the tag group can be classified into an approximate (ie, similar) class.

  Finally, the method proceeds to step 110. In step 110, the method ends.

  As described above, the overall flow of the method for classifying the tag groups according to the embodiment of the present invention has been described. Hereinafter, a specific flow of the classification step of the method for classifying the tag group will be described with reference to FIG. FIG. 2 is a flowchart illustrating a specific flow of a classification step of a method for performing classification on a tag group according to an embodiment of the present invention.

  As shown in FIG. 2, after calculating the similarity between the corresponding feature vector of the tag group and the corresponding core feature vector of each of the plurality of classes according to step 106, the method includes: Proceed to step 200. In step 200, the similarity between the calculated feature vector of the tag group and each core feature vector of at least one class is compared with a predetermined threshold. The predetermined threshold may be set in advance as necessary, or may be adjusted as necessary during the course of actual application. By adjusting the size of the threshold, it is possible to control the accuracy with which the tag group is classified.

  Here, it is assumed that there are three classes each constituted by a tag group and indicated by C1, C2, and C3. The corresponding core feature vectors of classes C1, C2, and C3 are A1, A2, and A3, respectively. When a new tag group TNE is found, the corresponding feature vector of the new tag group TNE is specified as ANE. The similarity between the feature vector ANE and the core feature vectors A1, A2, and A3 is calculated. For example, when the cosine similarity is adopted, the calculated similarity values are 0.92, 0.85, and 0.79, respectively. After calculating the similarity value, the similarity values, ie 0.92, 0.85 and 0.79, are respectively compared with a predetermined threshold.

  The method then proceeds to step 202. In step 202, it is determined whether the similarity between the calculated tag group and each class of at least one class exceeds a predetermined threshold value. If the determination result in step 202 is “NO”, that is, if the tag group is not similar to any of all classes, the process proceeds to step 206. In step 206, the tag group is included in the new class by classifying the tag group into a new class.

  In the above example, the predetermined threshold is 0.93. Since none of the calculated three similarity values 0.92, 0.85, and 0.79 exceeds a predetermined threshold, that is, 0.93, the new tag group TNE is similar to all of the current classes C1, C2, and C3. Not. At this time, a new tag group TNE can be included in the new C4 by generating a new class C4 and classifying the new tag group TNE into the new class C4.

  If the determination result in step 202 is “YES”, the process proceeds to step 204. In step 204, it is determined whether or not there are a plurality of corresponding classes having a degree of similarity greater than a predetermined threshold, that is, whether or not the degrees of similarity between the tag group and the plurality of classes are all greater than a predetermined threshold. Determine whether. If the determination result in step 204 is “NO”, it is determined that the similarity between the tag group and a certain class is greater than a predetermined threshold, that is, the number of similarities greater than the predetermined threshold is 1. Mean, go to step 210. In step 210, the tag group is classified into its corresponding class with a calculated similarity that exceeds a predetermined threshold.

  In the above example, the predetermined threshold value is 0.90. Of the calculated three similarity values 0.92, 0.85, and 0.79, only the similarity value 0.92 exceeds a predetermined threshold value 0.90. Therefore, a new tag group TNE is assigned a class corresponding to the similarity value 0.92. Classify as C1.

  If the determination result in step 204 is “YES”, it means that the degree of similarity between the tag group and the plurality of classes is larger than a predetermined threshold, that is, the number of similarities larger than the predetermined threshold is plural. Then, the process proceeds to Step 208. In step 208, the maximum similarity among a plurality of similarities greater than a predetermined threshold is selected, and the tag group is classified into that class corresponding to the selected maximum similarity.

  In the above example, the predetermined threshold is 0.80. Among the calculated three similarity values 0.92, 0.85, and 0.79, both the similarity values 0.92 and 0.85 exceeded the predetermined threshold value 0.80. Therefore, the similarity values 0.92 and 0.85 exceeding the predetermined threshold value 0.80. Among them, the maximum similarity value, that is, the similarity value 0.92 is selected. Thereafter, the new tag group TNE is classified into the class C1 corresponding to the maximum similarity value 0.92.

  After steps 206, 208 and 210, go to step 212. In step 212, the method ends.

  In the above description, the cosine similarity is used to calculate the similarity between the tag group and the tag group, and the similarity between the tag group and the class composed of the tag group. However, as long as the similarity between the tag groups or the tag group and the class composed of the tag groups can be calculated, other similarity calculation methods may be adopted. It should be appreciated by those skilled in the art.

  In the above description, the number of tag groups included in a class is dynamically increased. After the tag group is classified into a certain class by the method for classifying the tag group, the number of tag groups included in the class is increased by one. Preferably, after classifying a new tag group into a certain class, based on the new tag group and all tag groups already included in the class before that, the class using the formula (2) is used. Are newly calculated, and the newly calculated core feature vector is set as a new core feature vector corresponding to the class. Then, when classifying another tag group, the similarity between the another tag group and the new core feature vector of the class is compared. Therefore, according to the method of the present embodiment, it is possible to determine the same or similar between tag groups more accurately and more efficiently by comprehensively considering various features of various tag groups.

  Hereinafter, an apparatus for classifying tag groups according to an embodiment of the present invention will be described with reference to FIG. FIG. 3 is a block diagram illustrating an apparatus for classifying tag groups according to an embodiment of the present invention.

  As shown in FIG. 3, the apparatus 312 for classifying tag groups mainly includes a synonym tag set specifying unit 300, a feature vector generating unit 302, a similarity calculating unit 304, and a tag group classifying unit. 306. The synonym tag set specifying unit 300 specifies a synonym tag set to which each tag in the input tag group belongs based on a plurality of synonym tag sets stored in the synonym tag set database 308. The feature vector generation unit 302 generates a feature vector corresponding to the input tag group. In the generated feature vector, each element corresponds to a different synonymous tag set in a plurality of synonymous tag sets, and the value of each element is the number of tags belonging to the synonymous tag set corresponding to the element in the tag group. The similarity calculation unit 304 calculates the similarity between the feature vector and the core feature vector of each class of at least one class stored in the class set database 310. Here, the value of each element of the core feature vector of the class is the sum of the values of the corresponding elements in the corresponding vector of each tag group already classified into the class. The tag group classification unit 306 classifies the input tag group into an approximate class among at least one class stored in the class set database 310 based on the calculated similarity.

  The tag group classification unit 306 includes a class specifying unit 3062. The class identification unit 3062 determines whether each class of at least one class is based on whether the similarity between the calculated tag group and each class of the at least one class exceeds a predetermined threshold. Specifies whether or not is an approximate class. If there is no approximate class among at least one class, the class specifying unit 3062 classifies the tag group into a new class. If there are a plurality of approximate classes, the class specifying unit 3062 classifies the tag group into a class corresponding to the calculated maximum similarity.

  It should be understood by those skilled in the art that the plurality of synonymous tag sets may be provided in other ways, such as a synonym tag set dictionary, or the class may be provided in other ways. . The synonym tag set database 308 and the class set database 310 are stored in the storage unit 314. The storage unit 314 is, for example, a magnetic disk, flash memory, mobile memory, or the like. The storage unit 314 may be provided in the device 312 that performs classification on the tag group, or may be located outside the device 312 that performs classification on the tag group and may be wired or wireless. Means may be added to the device 312 for classifying the tag group.

  Using the cosine similarity, the similarity between the tag group and the tag group, and the similarity between the tag group and the class comprising the tag group can be calculated. However, if the similarity between the tag group and the tag group or the similarity between the tag group and the class comprising the tag group can be calculated, other similarity calculation methods may be adopted. Should be understood by those skilled in the art.

  The apparatus 312 for classifying the tag group is an apparatus corresponding to a method for actually classifying the tag group. Therefore, detailed description thereof is omitted here.

  Hereinafter, a method for mashing up data based on tag groups will be described with reference to FIG. FIG. 4 is a flowchart illustrating a method for mashing up data based on tag groups. This method is performed in the apparatus shown in FIG.

  As shown in FIG. 4, the method starts at step 400. The method then proceeds to step 402. In step 402, the classification unit 503 of the apparatus shown in FIG. 5 classifies the tag group into at least one class using the method for classifying the tag group. Therefore, by using the above-described classification method for tag groups, tag groups that meet different data format specifications, or different tag groups customized by the user, etc. can be dynamically changed according to the similarity between them. The tag groups in each class are similar to each other.

  The method then proceeds to step 404. In step 404, the replacement unit 505 of the apparatus shown in FIG. 5 replaces each tag of each tag group in the same class with the tag specified in the synonymous tag set to which it belongs. After the tag group is divided into different classes in step 402, each tag of each tag group in the same class can be replaced with a uniform tag. This unifies each tag that is similar in the same class into the same tag group, and uses the same tag group that has been acquired, and the data previously described in each similar tag group is updated. Mashup of data having similar contents and meaning can be realized.

  The replacement operation of each tag of each tag group in the same class can be performed by various methods. For example, each tag of each tag group in the same class can be replaced with a tag specified in the synonymous tag set to which the tag belongs. The designated tag may be, for example, the first tag or the last tag in the synonymous tag set to which each tag of each tag group in the same class belongs. Or, for example, for all tag groups in the same class, the usage frequency of each synonym tag in the synonym tag set to which each tag of each tag group in the same class belongs is statistically analyzed, and the synonym having the highest usage frequency is calculated. The tag may be the designated tag. For those skilled in the art, replacement of each tag of each tag group in the same class may be performed by other methods as long as it can be ensured that the specified tags after replacement can uniformly define corresponding data. Should be understood.

  The method then proceeds to step 404. In step 404, the method ends.

  An apparatus for mashing up data based on tag groups will be described with reference to FIG. FIG. 5 is a block diagram illustrating an apparatus for mashing up data based on tag groups.

  As shown in FIG. 5, the apparatus 501 for mashing up data based on tag groups mainly includes a classification unit 503 and a replacement unit 505. The classification unit 503 classifies the tag group in the input data into at least one class using an apparatus that classifies the tag group. The replacement unit 505 replaces each tag of each tag group in the same class with a tag specified in the synonymous tag set to which the tag group belongs. Thereby, it is possible to unify similar tags in the same class into the same tag group, and to newly describe the input data using the acquired same tag group, The mashup of the data which it has is implement | achieved.

  The device 501 for mashing data based on the tag group is actually a device corresponding to the method for mashing data based on the tag group. Therefore, detailed description thereof is omitted here.

  FIG. 6 is a block diagram illustrating an exemplary structure of a computer that implements the apparatus and method of the present invention.

  In FIG. 6, a central processing unit (CPU) 601 executes various processes based on a program stored in a read only memory (ROM) 602 or a program loaded from a storage unit 608 to a random access memory (RAM) 603. To do. In the RAM 603, data necessary when the CPU 601 executes various processes and the like is also stored as necessary.

  The CPU 601, ROM 602, and RAM 603 are connected to each other via a bus 604. An input / output interface 605 is also connected to the bus 604.

  An input unit 606 including a keyboard and a mouse, an output unit 607 including a display such as a cathode ray tube (CRT) and a liquid crystal display (LCD) and a speaker, a storage unit 608 including a hard disk, a LAN card, a modem, and the like The communication unit 609 including the network interface card is connected to the input / output interface 605. The communication unit 609 executes communication processing via a network, for example, the Internet.

  A drive 610 is also connected to the input / output interface 605 as necessary. A removable medium 611 such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory is attached to the drive 610 as necessary, and a computer program read from the medium 611 is stored as necessary. 608 is installed.

  When the steps and processes are realized by software, a program constituting the software is installed from a network such as the Internet or a storage medium such as a removable medium 611.

  Such a storage medium is not limited to a removable medium 611 in which a program is stored as shown in FIG. 6 and provided to the user by being distributed separately from the method. Should be understood. Examples of the removable medium 611 include a magnetic disk, an optical disk (including a compact disk read-only memory (CD-ROM) and a digital versatile disk (DVD)), a magneto-optical disk (including a mini disk (MD)), and a semiconductor memory. including. Alternatively, the storage medium may be a ROM 602, a hard disk included in the storage unit 608, or the like. The program is stored in it and distributed to users.

  The preferred embodiment of the present invention has been described above, but the present invention is not limited to this embodiment, and all modifications to the present invention belong to the technical scope of the present invention unless departing from the spirit of the present invention.

  The following additional notes are further disclosed with respect to the embodiment including the above examples.

  (Supplementary note 1) A method in which a computer classifies a tag group including at least one tag and corresponding data defined by the at least one tag, and the computer belongs to a synonymous tag From the synonymous tag set group and the tag group group to which the tag defining the data of one data list belongs, an element group indicating how many tags belonging to a certain group appear in which synonymous tag is generated, and the element group A feature vector corresponding to each tag group is generated, and the tag group is classified into classes according to the similarity of the feature vectors of each tag group.

  (Additional remark 2) The said computer calculates the core feature vector used as the sum of the value of the element of the feature vector of the tag group classified into the class about each class, the feature vector of the tag group which should be classified, and classification | category destination When the similarity with the core feature vector of each class is obtained and it is determined that it is not similar to any core feature vector, a new class is created and the tag group to be classified is classified into the created new class A method for performing classification according to appendix 1, wherein:

  (Appendix 3) A device for classifying a tag group including at least one tag and corresponding data defined by the at least one tag, the synonymous tag set group to which the synonymous tag belongs, From the tag group group to which the tags defining the data of one data list belong, a first unit for generating an element group indicating how many tags belonging to a certain group appear in which synonymous tag, and the element group, A second unit that generates a feature vector corresponding to each tag group; and a third unit that classifies the tag group into classes according to the similarity of the feature vectors of each tag group. A device that performs the classification.

  (Supplementary Note 4) For each class, a fourth unit that calculates a core feature vector that is the sum of the element vector element values of the tag group classified into the class, a feature vector of the tag group to be classified, and a classification When the similarity with the core feature vector of each previous class is obtained and it is determined that it is not similar to any of the core feature vectors, a new class is created and the tag group to be classified is changed to the created new class. An apparatus for performing classification according to supplementary note 3, including a fifth unit for classification.

  (Supplementary Note 5) A method in which a computer mashes data based on a tag group, and the computer classifies the tag group into at least one class by the method described in Supplementary Note 1 or 2, A method of mashing up data, wherein each tag of each tag group in the same class is replaced with a tag specified in the synonymous tag set to which the tag belongs.

  (Appendix 6) A device for mashing up data based on a tag group, the device for performing classification according to appendix 3 or 4, and a classification unit for classifying the tag group into at least one class, and in the same class An apparatus for mashing up data, comprising: a replacement unit that replaces each tag of each tag group with a tag specified in a synonymous tag set to which the tag group belongs.

  (Supplementary note 7) A program for classifying a tag group including at least one tag and corresponding data defined by the at least one tag, the synonymous tag set group to which the synonymous tag belongs. And a tag group group to which tags defining data of one data list belong, generate an element group indicating how many tags belonging to a certain group appear in which synonymous tag, and from the element group, each tag A program for generating a feature vector corresponding to a group and classifying the tag group into classes according to the similarity of the feature vectors of each tag group.

  (Supplementary note 8) A computer-readable storage medium storing the program according to supplementary note 7.

Claims (8)

A method wherein a computer classifies a tag group including at least one tag and corresponding data defined by the at least one tag,
The computer is
Generates an element group indicating how many tags belonging to a group appear in which synonym tag group from a synonym tag set group to which synonymous tags belong and a tag group group to which tags defining data of one data list belong And
Generating a feature vector corresponding to each tag group from the element group;
Classifying the tag groups into classes according to the similarity of the feature vectors of each tag group;
Here, in the generated feature vector, each element corresponds to a different synonym tag set in the synonym tag set group, and the value of each element corresponds to the synonym tag corresponding to the element in the corresponding tag group. A classification method characterized by the number of tags belonging to a set . The computer is
For each class, calculate a core feature vector that is the sum of the element vector element values of the tag groups classified into classes,
The similarity between the feature vector of the tag group to be classified and the core feature vector of each class to be classified is obtained, and if it is determined that it is not similar to any core feature vector, a new class is created and the classification is performed. Classify the tag group to be created into the newly created class,
The method according to claim 1, wherein the classification is performed. An apparatus for classifying a tag group including at least one tag and corresponding data defined by the at least one tag,
Generates an element group indicating how many tags belonging to a group appear in which synonym tag group from a synonym tag set group to which synonymous tags belong and a tag group group to which tags defining data of one data list belong The first unit to
A second unit for generating a feature vector corresponding to each tag group from the element group;
A third unit that classifies the tag groups into classes according to the similarity of the feature vectors of each tag group;
Only including,
Here, in the generated feature vector, each element corresponds to a different synonym tag set in the synonym tag set group, and the value of each element corresponds to the synonym tag corresponding to the element in the corresponding tag group. A device that performs classification characterized by the number of tags belonging to a set . For each class, a fourth unit for calculating a core feature vector that is the sum of the element vector element values of the tag groups classified into classes,
The similarity between the feature vector of the tag group to be classified and the core feature vector of each class to be classified is obtained, and if it is determined that it is not similar to any core feature vector, a new class is created and the classification is performed. A fifth unit for classifying the tag group to be assigned into the created new class;
The apparatus for performing classification according to claim 3, comprising: A computer mashup of data based on tag groups,
The computer is
The method for performing classification according to claim 1 or 2, wherein the tag group is classified into at least one class,
Replace each tag in each tag group in the same class with the tag specified in the synonymous tag set to which it belongs,
A method of mashing up data characterized by that. A device that mashes up data based on tag groups,
A classification unit according to claim 3 or 4, wherein the classification unit classifies the tag group into at least one class;
A replacement unit that replaces each tag of each tag group in the same class with a tag specified in the synonymous tag set to which the tag belongs;
A device for mashing up data characterized by including: A program for causing a computer to execute the classification method according to claim 1 or 2 .   A computer-readable storage medium storing the program according to claim 7.

Images