JP2020004157A - Classification method, apparatus, and program - Google Patents

Abstract

To improve the accuracy of classifying texts.SOLUTION: A receiving/analysis section 12 acquires multiple pieces of analysis result information each including a set of morphemes included in a text and attribute information of the morphemes, for any text out of received multiple texts. A dividing section 14 refers to a storage section storing morpheme information including a specific morpheme and attribute information of the specific morpheme, to determine whether a set of the specific morpheme included in the morpheme information and the attribute information of the specific morpheme is included in the acquired multiple pieces of analysis result information. When a result of determination is affirmative, the dividing section divides the text in a position, corresponding to an appearance position of a morpheme included in one piece of the analysis result information in any text, to generate multiple texts. A classification section 16 classifies the generated texts and other texts of the other received texts into multiple clusters.SELECTED DRAWING: Figure 9

Description

  The disclosed technology relates to a classification method, a classification device, and a classification program.

  2. Description of the Related Art Conventionally, documents (text data) described in a natural language are classified based on the described contents.

  For example, an information search system has been proposed in which questions and answers associated with each other are stored in a document storage unit, and answers are classified into clusters based on feature vectors of the respective answers in the document storage unit.

  Further, an FAQ candidate extraction system has been proposed in which discourse data and discourse semantics are input and a question sentence that is a FAQ candidate is extracted from the discourse data and output. In this system, the discourse semantics include flow information of each statement, and extract question / request statements uttered by the customer and set with a flow indicating a question or request sentence from the discourse data. Then, those containing the specified keyword are extracted from the question / request statement, the question / request statement is clustered, and the question / request statement representative of each cluster is output as an FAQ candidate.

  Further, an apparatus has been proposed that includes a viewpoint list storage unit that stores a viewpoint list including a tree-shaped viewpoint and attribute words, and a learning document information storage unit that stores a large number of learning document information related to each attribute word. . This apparatus extracts a plurality of keywords from shared content, and derives a first vector having the keywords as elements and the appearance frequency thereof as values. Then, for each keyword, for the learning text information in the attribute word that matches the keyword, a second vector having the word included in the learning text information as an element and the appearance frequency as a value is derived. Further, the similarity between the two vectors is calculated, a viewpoint list with similarity in which the similarities are associated is generated, and a viewpoint and an attribute word having the largest variance of the similarity are derived for each layer of the viewpoint list.

JP-A-2002-41573 JP 2012-3704 A JP 2012-70036 A

  However, for example, when each text contains a fixed expression, the fixed expression is not able to extract appropriate features from each document, and the classification of the document is appropriately performed. You may not be able to do it.

  In one aspect, the disclosed technology aims to improve text classification accuracy.

  As one aspect, the disclosed technology accepts a plurality of texts, and for each of the received texts, sets a set of a morpheme included in the text and attribute information of the morpheme. Get multiple analysis result information including Further, the disclosed technology refers to a storage unit that stores morpheme information including a specific morpheme and attribute information of the specific morpheme. Then, it is determined whether any one of the plurality of pieces of analysis result information obtained includes a set of the specific morpheme included in the morpheme information and attribute information of the specific morpheme. Make a decision. Further, according to the disclosed technology, when the determination result is positive, the one of the texts is divided at a position corresponding to the appearance position of the morpheme included in the one of the analysis result information in the one of the texts To generate multiple texts. Further, the disclosed technology classifies other texts among the received texts and the generated texts into clusters.

  As one aspect, there is an effect that the accuracy of text classification can be improved.

FIG. 4 is a diagram for explaining classification of documents. FIG. 4 is a diagram for explaining an influence of a fixed expression on a classification of a document. FIG. 4 is a diagram for explaining an influence of a fixed expression on a classification of a document. It is a figure for explaining classification of a text for extracting a fixed expression. FIG. 9 is a diagram for explaining a problem of text classification for extracting a fixed expression. It is a functional block diagram of a classification device concerning this embodiment. It is a figure showing an example of an analysis result by a reception analysis part. It is a figure showing an example of the dictionary for division. It is a figure for explaining division of a text. It is a figure showing an example of a word model. It is a figure showing an example of a classification result screen. It is a block diagram showing a schematic structure of a computer which functions as a classification device concerning this embodiment. It is a flowchart which shows an example of the classification processing in this embodiment. It is a flowchart which shows an example of a division process. 9 is a flowchart illustrating an example of a clustering process. 9 is a flowchart illustrating an example of a display control process.

  Hereinafter, an example of an embodiment according to the disclosed technology will be described with reference to the drawings.

  The classification device according to the present embodiment classifies each text included in a text set into a plurality of clusters in order to extract a fixed expression.

  Here, before describing the details of the embodiment, the reason for classifying the text in order to extract the fixed expression will be described. For example, consider a case in which documents such as e-mails for responding to incidents related to the system are classified, and the incident represented by each document is identified as to what kind of incident.

  For example, as shown in FIG. 1, a set of sentences in one incident is defined as one document, and each document included in a document set related to a plurality of incidents is included in a document by a method such as BoW (Bag of Words). Vectorize by frequency of word occurrence. Then, documents having high vector cosine similarity (“0.7”, “0.0”, “0.4” in FIG. 1) are grouped into clusters to classify the documents.

  Then, a characteristic word included in a document belonging to each cluster is extracted by TF (Term Frequency) -IDF (Inverse Document Frequency) or the like (a word indicated by an underline in FIG. 1), and is associated with each cluster, so that each cluster is extracted. It is possible to grasp what kind of incident the document contained in is. Note that the TF value and the IDF value are defined as follows.

TF value of word w = number of occurrences of word w in document / number of occurrences of all words in document IDF value of word w = log (total number of documents / number of documents including word w)

  However, in the above-described vectorization, the fixed expression included in each document is also vectorized, and the fixed expression becomes noise and affects the cosine similarity. For example, as shown in FIG. 2, question 1 is an inquiry about a server, and question 2 is an inquiry about a network, and the content of each question is different. However, since both of the question sentences include a fixed expression such as "Thank you, I am Kudo of the Service First Development Department", the cosine similarity will be high.

  Further, as shown in FIG. 3, when the document includes a fixed-form expression, the number of words appearing in the document increases, so that the TF value of a word to be a characteristic word decreases.

  In particular, in the field of responding to incidents related to the system, fixed expressions such as greetings and closing words tend to appear in sentences of inquiries by e-mail or the like from customers. In order to reduce the influence of the fixed expression on the classification of the document, the fixed expression may be deleted from each document. However, since the fixed expressions include proper nouns such as "I am Kudo of Service 1 Development Department" and unique expressions for each customer, it is difficult to define the fixed expressions in advance. It is.

  Therefore, for example, as shown in FIG. 4, each document included in the document set is divided at a location indicating a delimiter of one sentence such as “. (Period)” or “@n (line feed code)” included in the document. This creates a one-sentence text set. Then, it is conceivable that each sentence is vectorized and clustered, the text included in each cluster is confirmed, the cluster into which the fixed expression is classified is specified, and the fixed expression is extracted. Then, the extracted fixed expression is deleted from each document.

  In this way, by classifying the text and extracting the fixed expression, the fixed expression including the unique expression and the like can also be extracted. Therefore, in the present embodiment, the text is classified to extract the fixed expression.

  However, when one sentence created from each document is a compound sentence, the sentence is not regarded as a fixed expression as a whole, even though the fixed expression is partially included, and is not classified into a cluster into which the fixed expression is classified. There is. For example, as shown in FIG. 5, a sentence classified into cluster 2 includes a fixed expression “Tell me how to cope” similar to the fixed expression classified in cluster 1, but “unexpected sentence” Since the message has been output, the message is not classified into cluster 1 due to the influence of "".

  Therefore, in the present embodiment, texts are classified so that fixed expressions included in compound sentences can be extracted. Hereinafter, details of the present embodiment will be described.

  As shown in FIG. 6, the classification device 10 according to the present embodiment functionally includes a reception analysis unit 12, a division unit 14, a classification unit 16, and a display control unit 18. In a predetermined storage area of the classification device 10, a dictionary for division 22 and a word model 24 are stored. Note that the reception analysis unit 12 is an example of a reception unit and an acquisition unit of the disclosed technology, and the division unit 14 is an example of a determination unit and a generation unit of the disclosed technology.

  The reception analysis unit 12 receives a text set input to the classification device 10. For example, the reception analysis unit 12 includes a text such as an e-mail at the time of responding to an incident related to the system, sets a set of texts in one incident as one document, and formats each document included in the document set into a single text. Accept. The text of one sentence is formed, for example, by dividing the document at a portion indicating a break of one sentence such as “. (Period)” or “$ n (line feed code)” included in the document.

  The reception analysis unit 12 performs morphological analysis on each text included in the received text set, divides the text into morphemes, and adds attribute information such as part of speech and morpheme information to each morpheme. In addition, the reception analysis unit 12 performs a dependency analysis on each text using the result of the morphological analysis, and analyzes a dependency relationship for each phrase.

  FIG. 7 shows an example of the result of the morphological analysis and the result of the dependency analysis performed by the reception analysis unit 12 with respect to the text "An unexpected message has been output, so please tell me what to do about it." A in FIG. 7 is a morphological analysis result. In the example of FIG. 7, for each morpheme included in the text, a number is sequentially assigned from the beginning of the text, and the attribute information obtained as a result of the morphological analysis is associated with each morpheme. That is, the analysis result of the morphological analysis is a list of morphological information including each morpheme included in the text and attribute information of the morpheme. B in FIG. 7 is a dependency analysis result. In the example of FIG. 7, the text is divided into segments based on the result of the morphological analysis, and the dependency relationship between the segments is analyzed. Each clause is indicated by a box, and the dependency relationship between the segments is indicated by an arrow. ing.

  As shown in FIG. 8, the division dictionary 22 is a dictionary in which, when a text is divided, a morpheme information including a specific morpheme included in a phrase serving as a delimiter and attribute information of the specific morpheme is stored. It is. For example, morpheme information included in a clause that forms a predetermined adverb clause can be defined in the division dictionary 22 in advance.

  The division unit 14 refers to the division dictionary 22 and adds, for each text, the morpheme information included in the morphological analysis result by the reception analysis unit 12 to the specific morpheme included in the division dictionary 22 and the specific morpheme of the specific morpheme. It is determined whether a pair with the attribute information is included. When the determination result is affirmative, the dividing unit 14 divides each text at a position corresponding to the appearance position of a specific morpheme in the text.

  More specifically, as illustrated in FIG. 9, the dividing unit 14 specifies a phrase related to the phrase in order from the phrase at the end of the text, and adds the morpheme defined in the division dictionary 22 to the specified phrase. It is determined whether or not a set of a specific morpheme and attribute information that matches the set of the attribute information is included. In the example of FIG. 9, the phrase related to the last phrase (A in FIG. 9) includes “morpheme: purpose, part of speech: noun, morpheme information: adverbible” defined in the division dictionary 22 ( The text is divided immediately after this phrase, which is indicated by a broken-line circle in FIG. 9). For example, when an adverbial clause or the like is included in a text, the text is likely to be a compound sentence, and the text can be divided into a single sentence by dividing immediately after a clause including a morpheme representing the adverbial clause.

  In addition, the dividing unit 14 sets the first half of the divided text as a new text, and repeats the above processing from the last clause of the new text as shown in B in FIG. Thus, even a compound sentence including three or more contents can be divided into simple sentences.

  In addition, if the phrase relating to the phrase to be processed does not include morphological information that matches the morphological information defined in the dictionary for division 22, the dividing unit 14 does not divide the text into the phrase to be processed. Processing is continued from such a clause. If there is no phrase relating to the phrase to be processed, the dividing unit 14 advances the phrase to be processed by one to the beginning without dividing the text.

  By processing from the end, clauses related to predicates such as adverb clauses can be efficiently specified.

  The division unit 14 puts the text after the division into the single sentence set with the original text as the single sentence for the undivided text.

  Here, an example of the word model 24 is shown in FIG. The word model 24 includes a word vector table 24A and an IDF value table 24B. The word vector table 24A is a table in which words are associated with word vectors obtained by expressing the words by TF-IDF, word2vec, or the like. The IDF value table 24B is a table that stores a word and an IDF value of the word in association with each other. The IDF value may be generated in advance from an arbitrary document set, or may be generated from a document set that is the basis of an input text set.

  The classification unit 16 classifies the other texts included in the text set and the generated new texts, that is, each of the simple sentences included in the simple sentence set, into a plurality of clusters.

  More specifically, the classification unit 16 refers to the word vector table 24A included in the word model 24 and converts each simple sentence included in the simple sentence set into a vector. Also, the classifying unit 16 clusters each simple sentence by using a conventionally known clustering technique such as k-means or simple linking using the cosine similarity of the word vector of each simple sentence.

  In addition, the classification unit 16 extracts characteristic words from each of the plurality of clusters based on the appearance status of the words included in the simple sentence classified into each of the plurality of clusters, and associates the extracted characteristic words with each cluster. . Note that TF-IDF or the like can be used as the appearance status of the word. The characteristic word is an example of characteristic information and a representative morpheme of the disclosed technology.

  The display control unit 18 arranges the clusters in the index indicating the appearance frequency of the simple sentences included in each of the plurality of clusters based on the index regarding the appearance status of each simple sentence in the text set, and displays the clusters on the display device (shown in the drawing). (Omitted).

  For example, the display control unit 18 refers to the IDF value table 24B included in the word model 24, acquires the IDF value of each word included in each simple sentence, and calculates the norm of the IDF value vector of each simple sentence. Further, the display control unit 18 calculates, for each cluster, the average of the norms of the IDF value vectors of the simple sentences included in the cluster. Then, the display control unit 18 sorts the clusters in ascending order of the average of the norm of the IDF value vector, and displays the clusters on the display device. A cluster having a small average of the norm of the IDF value vector indicates that a simple sentence included in the cluster appears transversely in the text set. Therefore, the cluster is regarded as a cluster in which the fixed expression is classified. Things.

  FIG. 11 shows an example of the classification result screen 30 displayed on the display device. In the example of FIG. 11, each cluster is surrounded by one frame, and the simple sentences included in the cluster are displayed in the frame. Each cluster is displayed by the classifying unit 16 in association with a characteristic word associated with the cluster. FIG. 11 shows an example in which a cluster in which a fixed expression is classified is displayed at a higher level than a cluster in which a simple sentence indicating a specific content related to an incident such as a server or a network is classified.

  Note that the classification result screen 30 is not limited to the example of FIG. For example, a display form in which only the characteristic words associated with the cluster are displayed on the display device in the sort order, and the characteristic words are selected from the screen to display a simple sentence included in the cluster represented by the characteristic words. Good.

  The classification device 10 can be realized by, for example, a computer 40 illustrated in FIG. The computer 40 includes a CPU (Central Processing Unit) 41, a memory 42 as a temporary storage area, and a nonvolatile storage unit 43. Further, the computer 40 includes an input / output device 44 such as an input device and a display device, and an R / W (Read / Write) unit 45 that controls reading and writing of data from and to the storage medium 49. Further, the computer 40 includes a communication I / F 46 connected to a network such as the Internet. The CPU 41, the memory 42, the storage unit 43, the input / output device 44, the R / W unit 45, and the communication I / F 46 are connected to each other via a bus 47.

  The storage unit 43 can be realized by a hard disk drive (HDD), a solid state drive (SSD), a flash memory, or the like. The storage unit 43 as a storage medium stores a classification program 50 for causing the computer 40 to function as the classification device 10. The classification program 50 has a reception analysis process 52, a division process 54, a classification process 56, and a display control process 58. The storage unit 43 has an information storage area 60 in which information forming the division dictionary 22 and the word model 24 is stored.

  The CPU 41 reads out the classification program 50 from the storage unit 43, expands it in the memory 42, and sequentially executes the processes of the classification program 50. The CPU 41 operates as the reception analysis unit 12 illustrated in FIG. 6 by executing the reception analysis process 52. The CPU 41 operates as the dividing unit 14 illustrated in FIG. 6 by executing the dividing process 54. The CPU 41 operates as the classification unit 16 illustrated in FIG. 6 by executing the classification process 56. The CPU 41 operates as the display control unit 18 illustrated in FIG. 6 by executing the display control process 58. Further, the CPU 41 reads information from the information storage area 60 and expands the division dictionary 22 and the word model 24 in the memory 42. As a result, the computer 40 that has executed the classification program 50 functions as the classification device 10. Note that the CPU 41 that executes the program is hardware.

  The function realized by the classification program 50 can be realized by, for example, a semiconductor integrated circuit, more specifically, an ASIC (Application Specific Integrated Circuit).

  Next, the operation of the classification device 10 according to the present embodiment will be described.

  For example, a text set including a sentence such as an e-mail at the time of an incident response related to the system, and a set of sentences in one incident as one document, and forming each document included in the document set into one sentence text is sent to the classification device 10. Is entered. Then, the classification device 10 executes the classification processing shown in FIG. Note that the classification process is an example of a classification method of the disclosed technology.

  In step S10 of the classification process illustrated in FIG. 13, the reception analysis unit 12 receives the text set S input to the classification device 10. The text set S includes text 1, text 2,..., Text N (N is the number of texts included in the text set S).

  Next, in step S20, the division processing shown in FIG. 14 is performed.

  In step S22 of the dividing process shown in FIG. 14, an empty set is prepared as a single sentence set P, and in the next step S24, 1 is set to a variable s for identifying text.

  Next, in step S26, the reception analysis unit 12 performs a morphological analysis on the text s, divides the text into morphemes, adds attribute information to each morpheme, and adds a morpheme of each morpheme included in the text s. get information. In addition, the reception analysis unit 12 performs a dependency analysis on the text s using the result of the morphological analysis, and analyzes a dependency relationship for each phrase. .., M are assigned to the phrases of the text s in order from the beginning of the text s (m is the number assigned to the last phrase of the text s).

  Next, in step S28, the dividing unit 14 sets m to a variable i for identifying a clause of the text s.

  Next, in step S30, based on the result of the dependency analysis obtained in step S26, the dividing unit 14 determines whether or not a phrase j associated with the phrase i exists. If the clause j exists, the process proceeds to step S34. If the clause j does not exist, the process proceeds to step S32, and the dividing unit 14 sets the clause i-1, that is, the clause immediately before the clause i, as a new clause i, and the process proceeds to step S42. Move to.

  In step S34, the division unit 14 determines whether or not the phrase j associated with the phrase i includes morpheme information that matches the morpheme information defined in the division dictionary 22. If it is included, the process proceeds to step S36; otherwise, the process proceeds to step S40.

  In step S36, the dividing unit 14 divides the text s into a part s_1 after the clause j and a part s_2 before the clause j. Next, in step S38, the dividing unit 14 adds the part s_1 to the simple sentence set P and sets the part s_2 to a new text s. Next, in step S40, the dividing unit 14 sets the phrase j to a new phrase i.

  Next, in step S42, the division unit determines whether i is 0 or not, thereby determining whether the processing has been completed up to the beginning of the text s. If i = 0, the process proceeds to step S44, and if i has not yet reached 0, the process returns to step S30.

  In step S44, the dividing unit 14 adds the text s to the single sentence set P. As a result, the first half of the divided text is added to the simple sentence set P, and the original text of the undivided text is added to the simple sentence set P as it is.

  Next, in step S46, the reception analysis unit 12 determines whether or not s is N, thereby determining whether or not the processing in steps S26 to S44 has been completed for all the texts included in the received text set S. Is determined. If s has not reached N yet, the process proceeds to step S48, the reception analysis unit 12 increments s by 1, and the process returns to step S26. If s = N, the division process ends, and the process returns to the classification process.

  Next, the clustering process shown in FIG. 15 is executed in step S50 of the classification process shown in FIG.

  In step S52 of the clustering process illustrated in FIG. 15, the classification unit 16 vectorizes each simple sentence included in the simple sentence set P using the word vector table 24A of the word model 24.

  Next, in step S54, the classification unit 16 uses the cosine similarity of the word vector of each simple sentence or the like to cluster each simple sentence by a conventionally known clustering method such as k-means or simple linkage.

  Next, in step S56, the classification unit 16 determines a characteristic word from each of the plurality of clusters based on an index indicating the appearance state of a word included in the simple sentence classified into each of the plurality of clusters, such as TF-IDF. The extracted feature words are associated with each cluster. Then, the clustering process ends, and the process returns to the classification process.

  Next, the display control process shown in FIG. 16 is executed in step S60 of the classification process shown in FIG.

  In step S62 of the display control process shown in FIG. 16, the display control unit 18 refers to the IDF value table 24B included in the word model 24, acquires the IDF value of each word included in each simple sentence, and obtains the IDF value of each simple sentence. Calculate the norm of the IDF value vector.

  Next, in step S64, the display control unit 18 calculates, for each cluster, the average of the norms of the IDF value vectors of the simple sentences included in the cluster.

  Next, in step S66, the display control unit 18 sorts the clusters in ascending order of the average of the norm of the IDF value vector, and displays, for example, a classification result screen 30 as shown in FIG. 11 on the display device. Then, the display control processing ends, and the classification processing ends.

  As described above, according to the classification device of the present embodiment, each text included in a text set is divided immediately after a phrase including predetermined morpheme information such as an adverbial clause, and then the cosine of a word vector is divided. Clustering and classification are performed based on similarity and the like. Thereby, even when the text is a compound sentence and a part of the text includes a fixed expression, the classification accuracy of the text for extracting the fixed expression can be improved.

  Further, since the dependency relation is traced in order from the last clause included in the text to specify the clause including the predetermined morphological information such as the adverbial clause, it is possible to efficiently specify the division part.

  In the above-described embodiment, a case has been described in which clusters are sorted in ascending order of the average of the norms of the IDF value vectors of each simple sentence included in the cluster. However, the present invention is not limited to this. For example, clusters may be sorted in descending order of the number of simple sentences classified into each cluster. Since it is assumed that a fixed expression has a high appearance frequency in a text set, a cluster including many simple sentences is regarded as a cluster into which the fixed expression is classified.

  Further, in the above embodiment, it is assumed that the same morpheme defined in the division dictionary appears more than once in one text (for example, “for, for, please”). Even in such a case, the number assigned to each morpheme during the morphological analysis can specify which phrase contains which morpheme. Therefore, for example, even when one text includes the same morpheme, it is possible to identify those morphemes when determining a match with the morpheme information defined in the division dictionary.

  Further, in the above-described embodiment, a case has been described in which a text set obtained by shaping a document relating to the incident response of the system is input. However, the present invention is not limited to this, and the disclosed technology can be applied to various documents. In particular, the disclosed technique is effective for a document that includes many fixed expressions in the document.

  In the above embodiment, the mode in which the classification program is stored (installed) in the storage unit in advance has been described, but the present invention is not limited to this. The program according to the disclosed technology can be provided in a form stored in a storage medium such as a CD-ROM, a DVD-ROM, and a USB memory.

  Regarding the above embodiments, the following supplementary notes are further disclosed.

(Appendix 1)
Accept multiple texts,
For any one of the received texts, a plurality of pieces of analysis result information each including a set of a morpheme included in the text and attribute information of the morpheme are obtained,
With reference to a storage unit that stores morpheme information including a specific morpheme and attribute information of the specific morpheme, any one of the plurality of pieces of analysis result information obtained as analysis result information, The specific morpheme included is determined whether or not a set of attribute information of the specific morpheme is included,
When the determination result is positive, the one of the texts is divided at a position corresponding to the appearance position of the morpheme included in the one of the analysis result information in the one of the texts, and a plurality of texts are divided. Generate
Classifying the other texts of the received texts and the generated texts into a plurality of clusters,
A classification method, wherein the processing is performed by a computer.

(Appendix 2)
Dependency analysis of any of the texts to generate dependency information indicating a dependency relationship between a plurality of phrases included in any of the texts,
Based on the generated dependency information, specify a phrase having a specific dependency relationship with any one of the plurality of phrases,
Of the plurality of pieces of acquired analysis result information, a plurality of pieces of analysis result information each including a set of a morpheme included in the specified phrase and attribute information of the morpheme are specified,
When any of the analysis result information of the specified analysis result information includes a set of the specific morpheme included in the morpheme information and attribute information of the specific morpheme, the text of any one of Splitting to generate the plurality of texts,
The classification method according to Supplementary Note 1, wherein:

(Appendix 3)
Based on the text classified into each of the plurality of clusters, to generate feature information of each of the plurality of clusters,
Displaying the generated feature information on a display unit in association with each of the plurality of clusters;
The classification method according to Supplementary Note 1 or 2, wherein:

(Appendix 4)
The plurality of clusters respectively represent the plurality of clusters based on the text classified into each of the plurality of clusters and the appearance of the plurality of morphemes included in the text in the received plurality of texts. Determine a plurality of representative morphemes,
Displaying the determined plurality of representative morphemes on the display unit in association with the plurality of clusters represented by the plurality of representative morphemes,
The classification method according to appendix 3, characterized in that:

(Appendix 5)
Displaying each of the determined plurality of representative morphemes on the display unit in an order corresponding to the number of the texts classified into each of the plurality of clusters;
The classification method according to appendix 4, characterized in that:

(Appendix 6)
Based on the index related to the appearance status of each text in the plurality of texts, the index for the text included in each of the plurality of clusters is arranged on the display unit in the order indicating that the appearance frequency is high, and the clusters are displayed on the display unit. 6. The classification method according to any one of Supplementary Notes 3 to 5, wherein:

(Appendix 7)
A reception unit for receiving a plurality of texts,
An acquisition unit configured to acquire a plurality of pieces of analysis result information each including a set of a morpheme included in the text and attribute information of the morpheme for any of the plurality of texts received by the reception unit; When,
With reference to a storage unit that stores morpheme information including a specific morpheme and attribute information of the specific morpheme, the analysis result information of any of the plurality of pieces of analysis result information acquired by the acquisition unit A determination unit that determines whether a set of the specific morpheme included in the morpheme information and attribute information of the specific morpheme is included,
When the determination result by the determination unit is affirmative, the one of the texts is divided at a position corresponding to an appearance position of a morpheme included in the one of the analysis result information in the one of the texts, A generator for generating a plurality of texts,
A classification unit that classifies the other texts of the plurality of texts received by the reception unit and the plurality of texts generated by the generation unit into a plurality of clusters,
A classification device comprising:

(Appendix 8)
The acquisition unit performs dependency analysis on the text, and obtains dependency information indicating a dependency relationship between a plurality of phrases included in the text.
The determining unit, based on the dependency information obtained by the obtaining unit, among the plurality of phrases, to specify a phrase having a specific dependency relationship with any of the clauses, by the obtaining unit Of the plurality of pieces of acquired analysis result information, a plurality of pieces of analysis result information each including a set of a morpheme included in the specified phrase and attribute information of the morpheme,
The generation unit may include, in any analysis result information among the analysis result information specified by the determination unit, a set of the specific morpheme included in the morpheme information and attribute information of the specific morpheme. If included, split any of the texts to generate the plurality of texts;
8. The classification device according to claim 7, wherein

(Appendix 9)
The classifying unit generates feature information of each of the plurality of clusters based on the text classified into each of the plurality of clusters,
The display apparatus further includes a display control unit configured to display the feature information generated by the classification unit on a display unit in association with each of the plurality of clusters.
9. The classification device according to attachment 7 or 8, wherein:

(Appendix 10)
The classification unit, for the plurality of clusters, based on the text classified into each of the plurality of clusters, and a plurality of morphemes included in the text, the appearance status in the received plurality of texts, the plurality of Determine a plurality of representative morphemes each representing a cluster,
The display control unit displays each of the plurality of representative morphemes determined by the classification unit on the display unit in association with each of a plurality of clusters represented by each of the plurality of representative morphemes.
10. The classification device according to claim 9, wherein

(Appendix 11)
The display control unit, the plurality of representative morphemes determined by the classification unit, arranged on the display unit arranged in an order according to the number of the text classified into each of the plurality of clusters,
11. The classification device according to supplementary note 10, wherein:

(Appendix 12)
The display control unit, based on an index related to the appearance of each text in the plurality of texts, the index for the text included in each of the plurality of clusters, by arranging the clusters in the order indicating that the appearance frequency is high The classification device according to any one of Supplementary Notes 9 to 11, wherein the classification device is displayed on the display unit.

(Appendix 13)
Accept multiple texts,
For any one of the received texts, a plurality of pieces of analysis result information each including a set of a morpheme included in the text and attribute information of the morpheme are obtained,
With reference to a storage unit that stores morpheme information including a specific morpheme and attribute information of the specific morpheme, any one of the plurality of pieces of analysis result information obtained as analysis result information, The specific morpheme included is determined whether or not a set of attribute information of the specific morpheme is included,
When the determination result is positive, the one of the texts is divided at a position corresponding to the appearance position of the morpheme included in the one of the analysis result information in the one of the texts, and a plurality of texts are divided. Generate
Classifying the other texts of the received texts and the generated texts into a plurality of clusters,
A classification program for causing a computer to execute processing.

(Appendix 14)
Dependency analysis of any of the texts to generate dependency information indicating a dependency relationship between a plurality of phrases included in any of the texts,
Based on the generated dependency information, specify a phrase having a specific dependency relationship with any one of the plurality of phrases,
Of the plurality of pieces of acquired analysis result information, a plurality of pieces of analysis result information each including a set of a morpheme included in the specified phrase and attribute information of the morpheme are specified,
When any of the analysis result information of the specified analysis result information includes a set of the specific morpheme included in the morpheme information and attribute information of the specific morpheme, the text of any one of Splitting to generate the plurality of texts,
13. The classification program according to supplementary note 13, wherein

(Appendix 15)
Based on the text classified into each of the plurality of clusters, to generate feature information of each of the plurality of clusters,
Displaying the generated feature information on a display unit in association with each of the plurality of clusters;
13. The classification program according to Supplementary Note 13 or 14, wherein:

(Appendix 16)
The plurality of clusters respectively represent the plurality of clusters based on the text classified into each of the plurality of clusters and the appearance of the plurality of morphemes included in the text in the received plurality of texts. Determine a plurality of representative morphemes,
Displaying the determined plurality of representative morphemes on the display unit in association with the plurality of clusters represented by the plurality of representative morphemes,
16. The classification program according to supplementary note 15, wherein

(Appendix 17)
Displaying each of the determined plurality of representative morphemes on the display unit in an order corresponding to the number of the texts classified into each of the plurality of clusters;
18. The classification program according to supplementary note 16, wherein

(Appendix 18)
Based on the index related to the appearance status of each text in the plurality of texts, the index for the text included in each of the plurality of clusters is arranged on the display unit in the order indicating that the appearance frequency is high, and the clusters are displayed on the display unit. 18. The classification program according to any one of supplementary notes 15 to 17, characterized in that:

(Appendix 19)
Accept multiple texts,
For any one of the received texts, a plurality of pieces of analysis result information each including a set of a morpheme included in the text and attribute information of the morpheme are obtained,
With reference to a storage unit that stores morpheme information including a specific morpheme and attribute information of the specific morpheme, any one of the plurality of pieces of analysis result information obtained as analysis result information, The specific morpheme included is determined whether or not a set of attribute information of the specific morpheme is included,
When the determination result is positive, the one of the texts is divided at a position corresponding to the appearance position of the morpheme included in the one of the analysis result information in the one of the texts, and a plurality of texts are divided. Generate
Classifying the other texts of the received texts and the generated texts into a plurality of clusters,
A storage medium storing a classification program that causes a computer to execute processing.

10 Classification device 12 Reception analysis unit 14 Division unit 16 Classification unit 18 Display control unit 24 Word model 24A Word vector table 24B IDF value table 30 Classification result screen 40 Computer 41 CPU
42 memory 43 storage unit 49 storage medium 50 classification program

Claims (8)

Accept multiple texts,
For any one of the received texts, a plurality of pieces of analysis result information each including a set of a morpheme included in the text and attribute information of the morpheme are obtained,
With reference to a storage unit that stores morpheme information including a specific morpheme and attribute information of the specific morpheme, any one of the plurality of pieces of analysis result information obtained as analysis result information, The specific morpheme included is determined whether or not a set of attribute information of the specific morpheme is included,
When the determination result is positive, the one of the texts is divided at a position corresponding to the appearance position of the morpheme included in the one of the analysis result information in the one of the texts, and a plurality of texts are divided. Generate
Classifying the other texts of the received texts and the generated texts into a plurality of clusters,
A classification method, wherein the processing is performed by a computer. Dependency analysis of any of the texts to generate dependency information indicating a dependency relationship between a plurality of phrases included in any of the texts,
Based on the generated dependency information, specify a phrase having a specific dependency relationship with any one of the plurality of phrases,
Of the plurality of pieces of acquired analysis result information, a plurality of pieces of analysis result information each including a set of a morpheme included in the specified phrase and attribute information of the morpheme are specified,
When any of the analysis result information of the specified analysis result information includes a set of the specific morpheme included in the morpheme information and attribute information of the specific morpheme, the text of any one of Splitting to generate the plurality of texts,
The method according to claim 1, wherein: Based on the text classified into each of the plurality of clusters, to generate feature information of each of the plurality of clusters,
Displaying the generated feature information on a display unit in association with each of the plurality of clusters;
The classification method according to claim 1 or 2, wherein: The plurality of clusters respectively represent the plurality of clusters based on the text classified into each of the plurality of clusters and the appearance of the plurality of morphemes included in the text in the received plurality of texts. Determine a plurality of representative morphemes,
Displaying the determined plurality of representative morphemes on the display unit in association with the plurality of clusters represented by the plurality of representative morphemes,
4. The classification method according to claim 3, wherein: Displaying each of the determined plurality of representative morphemes on the display unit in an order corresponding to the number of the texts classified into each of the plurality of clusters;
The classification method according to claim 4, wherein: Based on the index related to the appearance status of each text in the plurality of texts, the index for the text included in each of the plurality of clusters is arranged on the display unit in the order indicating that the appearance frequency is high, and the clusters are displayed on the display unit. The classification method according to any one of claims 3 to 5, wherein: A reception unit for receiving a plurality of texts,
An acquisition unit configured to acquire a plurality of pieces of analysis result information each including a set of a morpheme included in the text and attribute information of the morpheme for any of the plurality of texts received by the reception unit; When,
With reference to a storage unit that stores morpheme information including a specific morpheme and attribute information of the specific morpheme, the analysis result information of any of the plurality of pieces of analysis result information acquired by the acquisition unit A determination unit that determines whether a set of the specific morpheme included in the morpheme information and attribute information of the specific morpheme is included,
When the determination result by the determination unit is affirmative, the one of the texts is divided at a position corresponding to an appearance position of a morpheme included in the one of the analysis result information in the one of the texts, A generator for generating a plurality of texts,
A classification unit that classifies the other texts of the plurality of texts received by the reception unit and the plurality of texts generated by the generation unit into a plurality of clusters,
A classification device comprising: Accept multiple texts,
For any one of the received texts, a plurality of pieces of analysis result information each including a set of a morpheme included in the text and attribute information of the morpheme are obtained,
With reference to a storage unit that stores morpheme information including a specific morpheme and attribute information of the specific morpheme, any one of the plurality of pieces of analysis result information obtained as analysis result information, The specific morpheme included is determined whether or not a set of attribute information of the specific morpheme is included,
When the determination result is positive, the one of the texts is divided at a position corresponding to the appearance position of the morpheme included in the one of the analysis result information in the one of the texts, and a plurality of texts are divided. Generate
Classifying the other texts of the received texts and the generated texts into a plurality of clusters,
A classification program for causing a computer to execute processing.