JP2018026119A - Classification system, control method of classification system, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To solve the problem in which work to give intracompany classification is not efficiently performed and feedback of a classification result cannot be efficiently utilized.SOLUTION: A classification system comprises a model construction unit that constructs a learning model based on teacher data which indicates a relationship between input data associated with at least one official report and output data indicating at least one piece of classification given to the at least one official report corresponding to the input data. The classification system inputs processing target data into the learning model, and provides information associated with the processing target data to a user terminal according to the result output from the learning model. When teacher information is received from the user terminal, the classification system submits the teacher data according to the teacher information to the model construction unit.SELECTED DRAWING: Figure 2

Description

  The present invention relates to a technique for classifying a gazette such as a patent gazette (including an application publication gazette and a patent publication gazette).

  The patent publication is assigned a classification number. For example, a patent gazette is issued after an international patent classification (IPC) is given. In Patent Document 1, when the user designates the international patent classification, the name of the invention of the patent publication to which the international patent classification to be designated and the summary are displayed, so that the user understands the international patent classification. Techniques that facilitate this are disclosed.

  The international patent classification is a classification commonly used in the world, but the user cannot arbitrarily specify the classification method. In order to facilitate patent searches even in R & D departments and technical departments that are unfamiliar with patent searches, proprietary classifications (referred to as in-house classifications) have been assigned to patent gazettes of companies and other companies.

JP 2007-233910 A

  As for the work of assigning the in-house classification, the person in charge of assigning the classification visually confirms, for example, a patent gazette issued in the new arrival and assigns the classification manually. Such work causes a load. In some cases, classification may be given by mistake. Further, although feedback indicating that the classification is not correct may be transmitted from the development department or the like to the person in charge of classification, it is not always reflected in the subsequent classification work.

  In addition, when the in-house classification is given, a qualitative evaluation may be given together with the target publication, and such an evaluation may be given by mistake. And such qualitative evaluation is not always performed at the same time as in-house classification, but may be performed independently.

  The classification system according to the present invention is based on teacher data indicating a relationship between input data relating to at least one publication and output data indicating at least one classification assigned to at least one publication corresponding to the input data. A model building unit that builds a learning model, processing target data related to at least one publication, and a processing target data input unit that inputs the learning model as the input data, and the learning according to the input of the processing target data According to the result output from the model, a providing unit that provides information related to the processing target data to a user terminal, a receiving unit that receives teacher information that is a source of the teacher data from the user terminal, and the teacher information A teacher data input unit that inputs the teacher data into the model construction unit.

  According to the present invention, it is possible to efficiently perform work for assigning in-house classifications. In addition, the feedback of the classification result can be used efficiently. Furthermore, according to the present invention, it is possible to efficiently perform the work of giving qualitative evaluation.

It is a figure explaining the outline | summary of a classification | category operation | work. It is a figure which shows an example of a structure containing a classification system. It is a figure which shows an example of a structure of a learning model. It is a figure explaining teacher data. It is a figure which shows an example of the flowchart of the process which a classification system performs. It is a figure which shows an example of UI screen in a user terminal. It is a figure which shows the example of the classification data with evaluation stored in classification completed DB. It is a figure explaining the value used for a conversion table. It is a figure explaining the other example of teacher data. It is a figure which shows the other example of UI screen in a user terminal. It is a figure explaining the other example of teacher data. It is a figure which shows the other example of UI screen in a user terminal. It is a figure which shows the other example of UI screen in a user terminal.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In addition, the structure demonstrated in the following embodiment is only an example, and this invention is not limited to the structure shown in figure.

<Usage classification usage pattern>
Prior to the description of the embodiment, first, a usage mode in which in-house classification is used will be described. FIG. 1 is a diagram illustrating an example of an in-house classification utilization mode. In-house classification is used in workflows such as patent search and patent information watching.

  In the first scene W1, the investigator sets a predetermined search formula and distribution destination. For example, a search condition used in an SDI (Selective Dissemination of Information) search and its distribution destination are set. The SDI search is a search using a search condition set in advance for a newly issued publication. The search result is automatically distributed to the set distribution destination. Since it is not necessary to classify all new publications, in the first scene W1, for example, new publications are roughly narrowed down using a predetermined search condition such as an international search classification. Here, “publication” includes various publications such as patent publications, utility models, designs, trademark publications, patent publication publications, and the like. It may include not only publications issued by the JPO, but also technical documents and technical papers that are periodically issued. Any object may be used as long as a classification is given to some document.

  In the second scene W2, the person in charge of classification gives an in-house classification to the new arrival publication. In-house classification can be freely set in the user company or the like. For example, it is possible to classify by development department or product. Hierarchical classification such as major classification, middle classification, and minor classification is also possible. Generally, in-house classification is used for sorting new publications. Of course, the sorting of new publications is an example, and in-house classification can be used in various forms. The new arrival bulletin given the in-house classification is sent to the corresponding department. Multiple in-house classifications may be given. For example, two in-house classifications, classification C1 and classification C2, may be given to a certain publication P1.

  In the third scene W3, the person in charge of each department confirms the new arrival bulletin sent. If the classification is wrong, the person in charge of classification is notified that the classification is wrong. Further, the person in charge of each department confirms the new arrival bulletin and assigns an evaluation (weighting) to each bulletin. For example, the user company or the like can uniquely set various evaluations (weights) such as a value indicating the degree of association with the company's technology or a value indicating the degree of the possibility of infringement.

  In the fourth scene W4, according to the evaluation (weighting) given by each department, the investigator watches the progress of the publication and discusses with the department in charge as necessary.

  The above is a description of the workflow for in-house classification that is generally performed.

<Embodiment 1>
In the present embodiment, a mode in which a classification system (to be described later) processes the classification work of the second scene W2 in the workflow shown in FIG. 1 will be described. Note that the classification system described in this embodiment does not have to be used in the workflow system as shown in FIG. 1, and can be used in any form.

<Configuration>
FIG. 2 is a diagram illustrating an example of the configuration of the classification system 200 according to the embodiment. The classification system 200 includes a model construction unit 201, a learning model 202, a teacher data input unit 203, a processing target data input unit 204, an output result provision unit 205, a provision destination database (DB) 206, a teacher information reception unit 207, and teacher information integration. A section 208, a teacher information reflection section 209, and a classified DB 210. FIG. 1 is merely an example of the configuration, and may include other configurations. Further, not all the configurations described in FIG. 1 are essential requirements. The classification system 200 may be configured to be able to communicate with the publication DB 250 through a network.

  The classification system 200 can be realized as an information processing apparatus. The information processing apparatus may include a CPU, a memory, an HDD, and a network interface. Each unit illustrated in FIG. 2 may function as each unit illustrated in FIG. 2 when a program stored in the HDD is temporarily read into the memory and the CPU executes the program read into the memory. . Further, at least a part of the units illustrated in FIG. 2 may be realized by a plurality of information processing apparatuses connected to each other through various networks. Further, at least a part (for example, the model construction unit 201 and the learning model 202) of each unit illustrated in FIG. 2 may be realized by distributed processing by a plurality of information processing apparatuses.

  The classification system 200 is configured to be able to communicate with a terminal used by a user (hereinafter referred to as a user terminal 261). The classification system 200 provides a search result to the user terminal 261, and will be described later from the user terminal 261. Teacher information may be received through a network.

  The user terminal 261 may be any type of terminal such as a personal computer, a tablet, or a mobile terminal. The classification system 200 manages login of users, and may provide the user terminal 261 with search conditions and search results unique to the logged-in user.

  The classification system 200 of the present embodiment may be configured to be different for each user company. This is because if company B, which is different from company A, uses a learning model 202 constructed based on teacher data customized for company A by company A, the desired result of company B cannot be obtained. Moreover, it is because it is against the intention of A company and it is not appropriate from the viewpoint of information protection to make A company and rival company C use the learning model 202 customized for A company by A company. Therefore, in the classification system 200 of this embodiment, the learning model 202 is customized for each user company. The user company is not limited to a company, and may be an arbitrary group or group.

  The model construction unit 201 constructs a learning model 202. In this specification, “construction” is used as a concept including generating a new model, changing or updating a generated model, and replacing an existing model with a newly generated model. And For example, the model construction unit 201 may newly generate a model when there is no learning model constructed. If there is a learning model that has already been constructed, the model construction unit 201 may construct a new learning model by changing or updating a part of the configuration of the learning model. The model construction unit 201 may newly generate a learning model and replace the learning model that has already been constructed with the learning model that has been created.

  The model construction unit 201 may construct a neural network as shown in FIG. 3 as the learning model 202, for example. The neural network includes an input layer 310, an intermediate layer 320, and an output layer 330. Each layer is composed of a plurality of nodes. In FIG. 3, the input layer 310 includes nodes 311, 312, and 313, the intermediate layer 320 includes nodes 321, 322, and 323, and the output layer 330 illustrates nodes 331, 332, and 333. . The number of nodes included in each layer is not limited to the illustrated example, and a form composed of a large number of nodes is common. The intermediate layer 320 is not limited to a single layer, and may be composed of a plurality of layers. A weight is set for each node, and a value obtained by multiplying the input data by the weight is transmitted to a subsequent node.

  In supervised learning, for example, input data and output data corresponding to the input data are prepared as teacher data. The model construction unit 201 repeats the process of adjusting the weight set to each node so that the input data becomes equal to the output data. In this way, the model construction unit 201 constructs the learning model 202. Thereafter, when data to be processed is input to the learning model 202, the learning model 202 outputs a result according to the constructed model.

  In the present embodiment, the model construction unit 201 may construct the learning model 202 by executing a known machine learning process. For example, the model construction unit 201 may use a CNN (Convolutional Neural Network) or an RNN (Recurrent Neural Network). Other methods may be used. The learning model 202 may be constructed not by a neural network but by SVM (Support Vector Machine). In this embodiment, the learning model 202 itself may be of any type.

  The teacher data input unit 203 inputs teacher data to the model construction unit 201. In the initial stage, the teacher data input unit 203 may input data obtained from the classified DB 210 and the publication DB 250 as teacher data.

  The publication DB 250 is a DB in which, for example, all available published publications are stored. As described above, the publication DB 250 includes various publications such as patent publications, utility models, designs, trademark publication publications, patent publication publications, etc., and is issued periodically as well as publications issued by the JPO. Technical literature and technical papers may be included. Thus, the publication DB 250 does not have to be a single DB, and may be a set of a plurality of DBs. The gazette DB 250 may include not only gazettes in Japan but also gazettes of gazettes in other countries.

  The classified DB 210 stores, for example, data in which identification information (for example, application number, publication number, etc.) of a publication is associated with an in-house classification assigned to the publication. For example, the publication P10 stores data indicating that the classification C1 and the classification C2 are given.

  The teacher data input unit 203 may input teacher data as shown in FIG. 4 as initial teacher data, for example. The teacher data may be data indicating a relationship between input data relating to at least one publication and output data indicating at least one classification given to at least one publication corresponding to the input data. More specifically, the output data may be data indicating the degree of approximation of each of a plurality of classifications that can be assigned as in-house classifications. FIG. 4 is a diagram illustrating an example of teacher data. In the example of FIG. 4, it is assumed that a plurality of classifications that can be given as in-house classifications are classification C1, classification C2, and classification C3. In the teacher data shown in FIG. 4, text data of a summary sentence of the publication P1 is prepared as input data, and the output data corresponding to the input data is a value 1 indicating the degree of approximation of the classification C1 with respect to the publication P1, and the publication P1. An example is shown in which data including a value 0 indicating the degree of approximation of the classification C2 and a value 1 indicating the degree of approximation of the classification C3 with respect to the publication P1 is prepared. In the present embodiment, the value 1 indicates the closest value. That is, the publication P1 prepares teacher data indicating that classification should be made into the classification C1 and the classification C3. On the other hand, in the present embodiment, the value 0 indicates a value that is not closest. That is, the gazette P1 prepares teacher data indicating that it should not be classified into the classification C2.

  Thus, the model construction unit 201 constructs the learning model 202 using the teacher data as shown in FIG. For example, the model construction unit 201 may change the weight of each node of the neural network constituting the learning model 202. When the publication P1 is input as processing target data to the learning model 202 constructed in this way, data indicating the respective approximation degrees of the classifications C1 and C2 is obtained as output. When the teacher data in FIG. 4 is used as initial teacher data, when the publication P1 is input as processing target data to the learning model 202, the respective approximation degrees of the classification C1 and the classification C3 are infinitely close to 1. can get. When another publication P11 is input to the learning model 202 as the processing target data, values indicating the degrees of approximation of the classification C1, the classification C2, and the classification C3 are respectively output.

  In FIG. 4, although the example using the text data of a summary sentence as input data was shown, it is not restricted to this. The text data of claims may be sufficient, and the text data of the whole sentence may be sufficient. In addition, the user may input an arbitrary text document. Bibliographic data may be input as input data, or image data may be input. Some of these data may be used, and a plurality of types of data may be used.

The teacher data input unit 203 may obtain necessary data corresponding to the input data (for example, text data of the summary sentence of the gazette) from the gazette DB 250 as initial teacher data, and necessary data corresponding to the output data ( For example, data indicating the classification given to a specific publication may be acquired from the classified DB 210. When necessary data corresponding to the input data is stored in the classified DB 210 (for example, in the example of FIG. 4, the summary sentence is also stored in the classified DB 210), the teacher data input unit 203 Teacher data obtained without referring to the DB 250 may be used.

  The teacher data may include additional information indicating which part of the publication (for example, bibliographic information, summary, claims, drawings, etc.) is used for learning. The processing target data input unit 204 may refer to the additional information and use a corresponding part in the publication input as processing target data as input data.

  The processing target data input unit 204 acquires new arrival information from the publication DB 250. The processing target data input unit 204 may acquire new arrival information at a predetermined timing. For example, new information may be acquired every time the publication DB 250 is updated. New arrival information may be acquired at predetermined time intervals from the previous acquisition timing of new arrival information, such as weekly and monthly. The new arrival information may be a newly issued gazette from the time when the processing target data input unit 204 acquires the previous arrival information. The processing target data input unit 204 inputs the publication specified by the new arrival information to the learning model 202 as processing target data. Note that the processing target data input unit 204 may use, as processing target data, a publication in which the target is narrowed down by predetermined search conditions from newly arrived information.

  Further, the processing target data input unit 204 may input an arbitrary publication other than the publication specified by the new arrival information to the learning model 202 as the processing target data. For example, when the number of in-house classifications is increased, already classified publications may be input to the learning model 202 as processing target data. In addition, when changing the filtering condition of newly arrived information due to expansion of the technical field or the like, published publications that have not been processed so far may be input to the learning model 202 as processing target data.

  The output result providing unit 205 receives identification information (for example, an application number, a publication number, or a patent number) for identifying the processing target data according to the result output from the learning model 202 when the processing target data is input to the learning model 202. You may provide to the user terminal 261.

  The output result providing unit 205 specifies the classification to be given to the publication of the processing target data based on the result output by the learning model 202. For example, the output result providing unit 205 specifies that the classification in which the value indicating the degree of approximation of each classification output from the learning model 202 exceeds a predetermined threshold is the classification given to the publication of the processing target data. Good. The output result providing unit 205 assigns the identified classification to the publication of the processing target data. The output result providing unit 205 may identify a user corresponding to the assigned classification, and provide identification information for identifying the processing target data to the user terminal 261 of the identified user.

  The providing destination DB 206 stores data indicating a providing destination user who provides information on a gazette to which a predetermined classification is assigned. For example, the user who provides the publication with the classification C1 may be data indicating that the user belongs to the development department D1. There may be a plurality of users to be provided. That is, if a plurality of users belong to the development department D1, it may be provided to the plurality of users. In addition, the provision destination may extend over a plurality of departments.

  The provided content may be bibliographic information such as identification information related to the gazette, and may include some information related to the content such as a summary. The output result providing unit 205 notifies the registered user's registered e-mail address and the like that the output result has been extracted. When there is an access from the user terminal 261 in response to this notification, the output result providing unit 205 may present the output result to the user terminal 261 that has accessed.

  Further, the output result providing unit 205 outputs, to the classified DB 210, data in which identification information of the gazette input as processing target data to the learning model 202 is associated with the classification indicated by the result output from the learning model 202. The classified DB 210 is updated. In this embodiment, the classified DB 210 is first updated using the classification indicated by the result output from the learning model 202. Then, the form which corrects the data stored in classified DB210 based on the teacher information from a user is demonstrated. However, the present invention is not limited to this example.

  The teacher information receiving unit 207 receives teacher information transmitted from the user terminal 261 in response to the output result provided from the output result providing unit 205. The teacher information may be information in which the user specifies whether the result output from the output result providing unit 205 (that is, the content output from the learning model 202) is correct (appropriate) or incorrect (appropriate). . For example, it is assumed that the publication P21 provided to the members of the development department D1 is a publication that is not related to the development department D1. In such a case, a member of the development department D1 notifies the teacher information that the publication is not related to the department. The teacher information receiving unit 207 receives such teacher information. In addition, when a member of the development department D1 can determine that the publication P21 is not related to the own department but rather to the development department D2 of another department, the teacher who designates that the classification P2 is newly assigned to the publication P21. Information may be received.

  The teacher information integration unit 208 integrates teacher information when the teacher information reception unit 207 receives a plurality of teacher information. There are two modes for the teacher information integration unit 208 to integrate.

  One is a case where a plurality of classifications are output from the learning model 202 for one publication. That is, the teacher information is sent from each of a plurality of classification destination user terminals. For example, it is assumed that the publication P3 is input to the learning model 202 as the processing target data, and the results corresponding to any of the classifications C1, C2, and C3 are output from the learning model 202. The teacher information receiving unit 207 received information indicating that the publication is not related to the own department as teacher information from members of the development department D1 corresponding to the classification C1 and members of the development department D2 corresponding to the classification C2. Assume that Then, the teacher information integration unit 208 integrates the teacher information received in this way. That is, the teacher information integration unit 208 integrates a plurality of teacher information, and integrates the teacher information with the content that the publication P3 is correct to output the classification C3. The gazette P3 may be integrated by further including teacher information indicating that it is not correct to output the classification C1 and the classification C2. The integrated teacher information is transmitted to the teacher information reflecting unit 209.

  Another case is a case where a plurality of users are registered in one classification providing destination. That is, it is a case where teacher information is received from each of a plurality of users in the same department. In such a case, there are various methods for employing teacher information. For example, the teacher information integration unit 208 may output only the teacher information that is shared by all users to the teacher information reflection unit 209. In other words, when some users make different evaluations, the teacher information integration unit 208 does not have to adopt the publication as teacher information. According to this configuration, teacher data in which the opinions of all the users are the same is reflected, so that a more robust learning model 202 can be constructed.

  The teacher information integration unit 208 may employ teacher information by majority vote when some users make different evaluations. When a large number of users perform evaluation, it is assumed that the evaluations of all users do not completely match. If there is little teacher information that can be adopted, teacher data will not be a sufficient number of samples. As a result, there is a case where the learning model 202 does not advance and a sufficient result cannot be obtained. Therefore, when some users make different evaluations, the teacher information integration unit 208 may employ teacher information by majority vote.

  The teacher information integration unit 208 may change the weight for each user. The proficiency level varies from user to user. Therefore, the weight is increased for users with a high level of proficiency, and the weight is reduced for users with a low level of proficiency. Then, a result obtained by weighting and averaging the values evaluated by the respective users (that is, teacher information) may be adopted as the teacher information output to the teacher information reflecting unit 209. For example, even if a plurality of users with low proficiency levels output teacher information indicating that classification is inappropriate, if users with high proficiency levels do not determine that classification is inappropriate, teacher information The integration unit 208 may not output the teacher information that the classification is inappropriate to the teacher information reflection unit 209.

  The teacher information integration unit 208 may be configured to accept an instruction from a user as a leader. When some users among the plurality of users make different evaluations, it may be determined whether or not to adopt the teacher information according to an instruction from the reader. Specifically, the teacher information integration unit 208 determines whether teacher information different from others is included. When different teacher information is included, a UI screen for inquiring which teacher information to use is provided to a user terminal registered as a reader. When receiving the designation of teacher information to be adopted from the user terminal, the teacher information integration unit 208 adopts the designated teacher information.

  The teacher information reflection unit 209 reflects the teacher information output from the teacher information integration unit 208. The teacher information reflection unit 209 updates the classified DB 210 based on the teacher information. For example, when the publication P3 is input to the learning model 202 as the processing target data as in the above example, the degree of approximation of the classification C1 output from the learning model 202 is 0.6, and the degree of approximation of the classification C2 is 0. 7. A case where the degree of approximation of classification C3 is 0.8 will be described as an example. In response to this result, the output result providing unit 205 may determine that a classification having an approximation degree exceeding a predetermined threshold (for example, 0.5) is an appropriate classification. As a result, the output result providing unit 205 specifies that the classifications corresponding to the publication P3 are the classification C1, the classification C2, and the classification C3. And the information regarding the publication P3 is output to the user terminal 261 corresponding to each classification. Further, the output result providing unit 205 stores the classified data in which the publication P3 is associated with the classifications C1, C2, and C3 in the classified DB 210. At this time, it is assumed that the integrated teacher information is information indicating that classification C3 is appropriate and classification C1 and classification C2 are inappropriate for the publication P3. Then, the teacher information reflection unit 209 changes the classification data regarding the publication P3 that is already stored in the classified DB 210. That is, the already stored classification data relating to the publication P3 is changed to classification data in which the publication P3 is associated with only the classification C3.

  Also, the teacher information reflection unit 209 outputs teacher data based on the integrated teacher information to the teacher data input unit 203. In other words, the gazette P3 is input data, and the teacher data input unit 203 outputs the teacher data whose output data is the approximation value 0 of the classification C1, the approximation value 0 of the classification C2, and the approximation value 1 of the classification C3. To do. In response to this, the teacher data input unit 203 inputs the teacher data to the model construction unit 201 and constructs the learning model 202 again. By this processing, the subsequent processing target data is processed using the learning model reflecting the teacher data.

  FIG. 5 is a flowchart illustrating an example of processing according to the present embodiment. The process shown in FIG. 5 is executed in the classification system 200 when a user logs in to the classification system 200 using the user terminal 261 and constructs a learning model using teacher data.

  In step S501, the teacher data input unit 203 acquires teacher data. When a new learning model is constructed, the teacher data input unit 203 refers to the classified DB 210 and acquires classified data that has been classified manually by a classifier in the past as teacher data. You can do it.

  In step S <b> 502, the teacher data input unit 203 inputs the acquired teacher data into the model construction unit 201. The model construction unit 201 constructs a learning model 202 using the input teacher data. The initial teacher data may be arbitrary teacher data, and is not limited to the example described. Further, in the state where the learning model has already been constructed, the processing in step S501 and step S502 may be omitted.

  In step S <b> 503, the processing target data input unit 204 acquires processing target data from the publication DB 250. For example, the processing target data input unit 204 acquires new arrival information related to a gazette issued during a predetermined period. The processing target data input unit 204 may acquire new arrival information according to various conditions such as the distribution frequency set by the user terminal 261. The processing target data input unit 204 inputs the acquired new arrival publication as processing target data to the learning model 202. The learning model 202 outputs a result based on the input processing target data.

  In step S504, the output result providing unit 205 provides the user terminal with information regarding the processing target data according to the output result. The learning model 202 outputs data indicating the degree of approximation of each classification corresponding to the input data publication. The output result providing unit 205 identifies a classification having an approximation degree exceeding a predetermined threshold. The output result provision unit 205 refers to the provision destination DB 206 and identifies a provision destination user corresponding to the identified classification. Then, the result is provided to the user terminal of the identified user. For example, the output result providing unit 205 may provide the user terminal with a list of identification information for identifying the processing target data.

  In step S505, the teacher information receiving unit 207 receives teacher information from the user. The user terminal 261 receives, for example, gazette data with a classification related to the development department to which the user belongs, and a list of gazettes. The user confirms the publication using the user terminal 261, and designates that the publication that has been determined to be unrelated to its own development department is an appropriately classified publication.

  FIG. 6 is a diagram illustrating an example of a UI screen 600 for designating that a gazette that the user has determined to be unrelated to his / her development department is a gazette that is inappropriately classified. The UI screen 600 is displayed on the user terminal 261. On the UI screen 600, a list of newly published publications for the development department D1 to which the user belongs is displayed. The user can check the check boxes displayed side by side in the list, specify that the publication is classified inappropriately, or specify an appropriate classification destination using a pull-down menu. FIG. 6 shows an example of a list screen, but check boxes and pull-down menus may be displayed on a detail screen displaying details of each publication. When it is designated via the UI screen 600 that the publication is classified inappropriately, the information is transmitted as teacher information. When the user can determine an appropriate classification destination, an appropriate classification destination may be designated by a pull-down menu. In this case, the appropriate classification destination may also be transmitted as teacher information.

  In step S506, the teacher information integration unit 208 integrates a plurality of teacher information received by the teacher information reception unit 207. For example, the teacher information integration unit 208 integrates the teacher information when a plurality of teacher information is received by the teacher information reception unit 207 regarding a specific publication.

  In step S507, the teacher information reflection unit 209 reflects the teacher information in the classified DB 210. For example, in the classified DB 210, it is assumed that classification data whose classifications are classification C1, classification C2, and classification C3 are stored. At this time, when teacher information indicating that the classification C1 and the classification C2 are wrong is obtained, the teacher information reflection unit 209 updates the classification data associated with the publication P3 to classification data having only C3.

  In step S508, the teacher information reflection unit 209 generates teacher data based on the teacher information. In the above example, the gazette P3 is used as input data, and teacher data is generated with output data of the approximation value 0 of the classification C1, the approximation value 0 of the classification C2, and the approximation value 1 of the classification C3. The generated teacher data is output to the teacher data input unit 203.

  In step S509, the teacher data input unit 203 outputs the teacher data to the model construction unit 201. The model construction unit 201 constructs a learning model 202 based on teacher data.

  In step S510, the processing target data input unit 204 determines whether there is next processing target data. For example, it is determined whether it is time to acquire the next new publication. If there is next data to be processed, the process proceeds to step S503, and the process is repeated thereafter.

  Although not shown, the processing shown in FIG. 5 may be ended when an instruction to stop the classification processing is received from the user terminal 261.

  As described above, according to the present embodiment, the process of assigning in-house classification can be executed by the learning model 202, so that classification can be efficiently performed. Further, by using the result output from the learning model 202 as teacher data, the learning model 202 can be evolved, and the feedback of the classification result can be used efficiently.

<Embodiment 2>
In the first embodiment, the description has been given of the mode in which, when information related to a specific publication is input as processing target data, the learning model outputs the degree of approximation of each classification for the specific publication. As described in the section “Usage classification usage pattern”, each development department evaluates each classified publication. In the second embodiment, a mode in which evaluation is used as a parameter for adjusting the degree of approximation will be described. In the example of the first embodiment, the example in which the value 1 indicating that the classification is appropriate and the value 0 indicating whether it is not appropriate has been described as the teacher data. In the second embodiment, a mode will be described in which the value of output data used as teacher data is used as an arbitrary value between 0 and 1 according to the evaluation.

  In the first embodiment, the value indicating the degree of approximation output from the learning model 202 has been described as being used in a scene for determining whether the classification is correct. In the second embodiment, an embodiment in which the order of the list of information provided to the user terminal is changed according to the value indicating the degree of approximation will be described.

  The configuration of the classification system in the second embodiment can be the same as that described in the first embodiment.

  In the second embodiment, the classified DB 210 stores classified data with evaluation in which at least one classification and an evaluation of the classification are associated with each publication. FIG. 7 is a diagram illustrating an example of classified data with evaluation stored in the classified DB 210. For example, in the publication P31, two classifications C1 and C2 are associated. Further, the classification C1 of the publication P31 is associated with an evaluation of high weighting, and the classification C2 of the publication P31 is associated with an evaluation of small weighting. Evaluations such as large weighting and small weighting are evaluations given in each development department. For example, an evaluation indicating the degree of relevance with the company's technology. The higher the evaluation, that is, the higher the weighting, the higher the degree of association with the company's technology, indicating that the publication is worthy of attention.

  In the second embodiment, the teacher data input unit 203 uses classification data with an evaluation as teacher data. The teacher data input unit 203 scores the classified data with evaluation as shown in FIG.

  FIG. 8 is a diagram for explaining an example of scoring. FIG. 8 is a diagram illustrating an example of a conversion table used when converting into a score. In the conversion table, weighting parameters for each evaluation are set. For example, when the evaluation attached to the classification C1 of the publication P31 is “high weighting”, a score of value 1 is attached to the classification C1 of the publication P31. When the evaluation assigned to the classification C2 of the publication P31 is “low weighting”, the classification C2 of the publication P31 is assigned a score of 0.6.

  FIG. 9 is a diagram illustrating an example of teacher data according to the second embodiment. As described in the first embodiment, the text data of the summary sentence is used as input data for each publication. On the other hand, unlike Embodiment 1, as output data, a value scored based on evaluation is used as a value indicating the degree of approximation. FIG. 9 shows an example in which the publications P31 and P32 shown in FIG. 7 are used as teacher data. In the publication P31, the classification C1 and the classification C2 are given, and the classification C3 is not given. Therefore, in the teacher data of FIG. 9, the value 0 is set as the degree of approximation of the classification C3 with respect to the publication P31. Further, since the weight assigned to the classification C1 of the publication P31 is “high weighting”, the value of 1 is set as the degree of approximation of the classification C1 with respect to the publication P31 in the teacher data of FIG. Further, since the weight assigned to the classification C2 of the publication P31 is “low weighting”, the degree of approximation of the classification C3 with respect to the publication P31 is set to a value of 0.6 in the teacher data of FIG. .

  When the teacher data input unit 203 inputs such teacher data into the model construction unit 201, the model construction unit 201 constructs a learning model 202 according to the teacher data. When the processing target data input unit 204 inputs processing target data to the constructed learning model 202, the learning model 202 outputs a predetermined output value (a value indicating the degree of approximation) as output data.

  The output result providing unit 205 provides the user terminal 261 with the result output from the learning model 202. For example, as described in the first exemplary embodiment, the output result providing unit 205 identifies a classification having a degree of approximation exceeding a predetermined threshold as a classification of a publication that is input as processing target data. The output result providing unit 205 may provide the user terminal 261 with a list of publications in which the order of arrangement is changed according to the degree of approximation among publications having the degree of approximation exceeding a predetermined threshold. For example, it is assumed that a list of publications classified as classification C1 is provided to the user of the development department D1. Further, it is assumed that a new publication group is input to the learning model 202 as processing target data. In this case, the output result providing unit 205 extracts a gazette that is output from the learning model 202 and has an approximate degree related to the classification C1 that exceeds a predetermined threshold. Then, a list in which the order of publications is changed in descending order of the value of the degree of approximation with respect to the classification C1 is generated from the extracted publications. The output result providing unit 205 may provide the user terminal 261 with the list generated in this way.

  FIG. 10 is a diagram illustrating an example of a UI screen for displaying the list provided from the output result providing unit 205 on the user terminal 261. In the second embodiment, it is possible to give weights to the respective publications. The gazette thus weighted is received by the teacher information receiving unit 207 as teacher information. Here, the list box is used for assigning the weight, but it may be an inbox for directly inputting a value or a control for inputting another screen.

  As described in Embodiment 1, the teacher information integration unit 208 may integrate a plurality of teacher information. For example, majority voting may be employed, weighting may be employed, or user deciding may be employed.

  As described in the first embodiment, the teacher information reflecting unit 209 generates teacher data reflecting the teacher information, and outputs the teacher data to the teacher data input unit 203. For example, teacher data reflecting weights is generated using a conversion table as shown in FIG. In the present embodiment, an example in which a conversion table is common to all classifications is shown, but the present invention is not limited to this. A different conversion table may be used for each classification. The teacher data input unit 203 inputs teacher data into the model construction unit 201, and the model construction unit 201 constructs a learning model 202.

  As described above, in this embodiment, since the user's evaluation can be included in the elements of the teacher data, the learning model 202 that outputs a more appropriate classification can be constructed.

  In the second embodiment, in order to simplify the description, an example in which only the major classification is used has been described. However, the middle classification may be included in the major classification, and a configuration in which a minor classification is further provided in the middle classification. Good. In such a case, the parameters may be further subdivided.

  In the second embodiment, the example in which the order of publications is changed and displayed on the user terminal 261 in descending order of the degree of approximation has been described, but the same display mode may be adopted in the first embodiment. That is, even when no evaluation (weighting) is given to the classification and information regarding the evaluation (weighting) is not used as the teacher information, the arrangement order is changed according to the degree of approximation output by the learning model 202. A list of publications may be provided to the user terminal 261.

<Embodiment 3>
In the first embodiment, the description has been given of the mode in which, when information related to a specific publication is input as processing target data, the learning model outputs the degree of approximation of each classification for the specific publication. In the third embodiment, a description will be given of a mode in which the learning model outputs the degree of risk for a specific publication when information on the specific publication is input as processing target data in addition to the classification approximation. Here, the risk level means incompatibility of the target company, target business, target product, target part, etc., and more specifically, whether the target object conflicts with a specific publication. It is a degree to show. In the third embodiment, first, in addition to the classification approximation described in the first embodiment, the learning model 202 is made to learn the risk, the processing target data is input to the learning model, and the classification approximation and the risk are output. An example is given.

  The configuration of the classification system in the third embodiment can be the same as that described in the first embodiment.

  In the third embodiment, the classified DB 210 stores at least one classification, an approximation degree of the classification, and a risk level in association with each publication.

  In the third embodiment, the teacher data input unit 203 uses, as teacher data, a gazette stored in the classified DB 210, a classification given to this gazette, an approximation degree thereof, and a risk degree.

  FIG. 11 is a diagram illustrating an example of teacher data according to the third embodiment. As described in the first embodiment, the text data of the summary sentence is used as input data for each publication. FIG. 11 shows an example in which the publication P1 and the publication P2 are used as teacher data. In the publication P1, the degree of approximation 1 is assigned to the classification C1, the degree of approximation 0 is assigned to the classification C2, the degree of approximation 1 is assigned to the classification C3, and the degree of risk is assigned 0.8. Further, in the publication P2, the degree of approximation 1 is assigned to the classification C1, the degree of approximation 0 is assigned to the classification C2, the degree of approximation 0 is assigned to the classification C3, and the degree of risk 1 is assigned. Here, in this example, one risk is assigned to one publication, but a risk may be assigned to one classification of one publication.

  When the teacher data input unit 203 inputs such teacher data into the model construction unit 201, the model construction unit 201 constructs a learning model 202 according to the teacher data. When the processing target data input unit 204 inputs processing target data to the constructed learning model 202, the learning model 202 outputs a predetermined output value (classification approximation and risk) as output data.

  The output result providing unit 205 provides the user terminal 261 with the result output from the learning model 202. For example, as described in the first exemplary embodiment, the output result providing unit 205 identifies a classification having a degree of approximation exceeding a predetermined threshold as a classification of a publication that is input as processing target data. The output result providing unit 205 may provide the user terminal 261 with the degree of risk in a list of publications in which the order of arrangement is changed according to the degree of approximation among publications having the degree of approximation exceeding a predetermined threshold. For example, it is assumed that a list of publications classified as classification C1 is provided to the user of the development department D1. Further, it is assumed that a new publication group is input to the learning model 202 as processing target data. In this case, the output result providing unit 205 extracts a gazette that is output from the learning model 202 and has an approximate degree related to the classification C1 that exceeds a predetermined threshold. Then, a list in which the order of publications is changed in descending order of the value of the degree of approximation with respect to the classification C1 is generated from the extracted publications. The output result providing unit 205 may provide the user terminal 261 with the list generated in this way. Here, since the risk level is also included in the list, it is possible to generate and use a list in which the order of publications is changed in descending order of the risk value. Further, the publication presented to the user may be a publication in which the degree of approximation does not exceed a predetermined threshold, and instead, the publication in which the degree of risk exceeds a predetermined threshold.

  FIG. 12 is a diagram illustrating an example of a UI screen for displaying the list provided from the output result providing unit 205 on the user terminal 261. In the third embodiment, the classification and the degree of risk are assigned to each publication and displayed. Here, the degree of approximation can also be displayed in the displayed list. The user can also correct any of the classification and the degree of approximation thereof, and the degree of risk. The subject publication, the corrected classification and the degree of approximation thereof, and the degree of danger are teacher information as teacher information. Received by the receiving unit 207.

  As described in Embodiment 1, the teacher information integration unit 208 may integrate a plurality of teacher information. For example, majority voting may be employed, weighting may be employed, or user deciding may be employed.

  As described above, in the present embodiment, since the user's evaluation can be included in the elements of the teacher data, the learning model 202 that outputs more appropriate classification and risk can be constructed.

  In the third embodiment, the learning model 202, which is one learning model, has been described as a configuration in which the gazette, classification, approximation, and risk are learned as teacher data. These learning models may be configured separately, and the learning model for the degree of risk may be configured by further subdividing the learning model by classification. Specifically, when there is a classification of C1 and C2, there is one learning model for classification, and a learning model for risk is prepared for C1 and C2. In the case of such a configuration, the classification, the degree of approximation, and the degree of risk preset by the user are read out, and first, the subject publication, the classification and the degree of approximation are learned in a learning model for classification, and then the classification is performed. The published gazette is learned using the gazette and the risk level in a learning model for the risk level corresponding to the classification. When information related to a specific publication is input as processing target data, first, the learning model for classification outputs an appropriate classification and is processed by the learning model for the risk corresponding to the output classification. The risk level of the subject publication is output. The degree of risk here can be said to be the degree of risk in the target classification.

  In addition, in order to output the risk level in the target classification described above, the learning model for the risk corresponding to the classification is used in addition to the learning model for the classification. It may be configured to output the risk level of the target department and the risk level of the target device, which are realized by building a learning model for the risk level of the target department and a learning model for the risk level of the target device, respectively. be able to. In this case, the teaching data to be input can cause the publication data and the risk level of the publication data to be learned by the learning model for the risk level of the corresponding target department and the learning model for the risk level of the corresponding target device. Necessary. Thereby, for example, the user discovers a plurality of publications that may be in conflict with the product A, assigns a value obtained by quantifying the risk level, and causes the learning model for the risk level of the product A to learn a new one. Is input as processing target data to this learning model, the learning model outputs the degree of risk for the target publication. That is, the degree of risk that the product A is in conflict with the target publication is output. In addition, in the embodiment, the description has been made on the assumption that the classification is given by the learning model, but the configuration that gives the risk level by the learning model described above does not necessarily include the operation of giving the classification, Only the operation of outputting the risk level of the subject publication by the learning model is included in the scope of the present invention. On the contrary, the structure which calculates | requires a danger level with a learning model and gives a classification | category after that may be sufficient.

  Furthermore, although the risk level is explained in the third embodiment, the development department may set the risk level instead of weighting in W3 of FIG.

  In the above-described embodiment, the description has been mainly given of the classification of publications relating to patents. It can be applied to classification processing of any kind of publication.

<Embodiment 4>
In the first embodiment, a description has been given of a mode in which, when information related to a specific gazette is input as processing target data, the learning model outputs a degree of approximation of each of a plurality of in-house classifications that can be assigned to the specific gazette. In the fourth embodiment, when information on an application is input as processing target data, the learning model is IPC, CPC (Cooperative Patent Classification), ECLA (European Classification), ICO (In Computer Only), FI (File Index), A description will be given of a classification by a plurality of domestic and foreign patent offices and international organizations that can be given to the application, such as F-term, and a form for outputting the degree of approximation.

  The configuration of the classification system in the fourth embodiment can be the same as that described in the first embodiment.

  The model construction unit 201 of the classification system 200 according to the present embodiment generates a learning model based on teacher data indicating a relationship between a plurality of input data related to an application and output data indicating at least one classification assigned to the application. To construct. The process executed by the model construction unit 201 is the same as that described in the first embodiment. Here, the application may be a patent application, a utility model registration application, a design registration application, or a trademark registration application, and the plurality of input data relating to the application is data relating to documents and properties submitted at the time of application.

  The processing target data input unit 204 inputs a plurality of processing target data related to the application into the learning model as a plurality of input data. More specifically, when the application is a patent application or utility model registration application, the plurality of data to be processed includes at least one of an application, a description, a claim or a utility model registration request, a drawing, and an abstract. Either may be included. In addition, when the application is a design registration application, the plurality of processing target data includes at least one of the application, the article related to the design described in the application, the description of the article related to the design, the description of the design, and the drawings. Good. Here, the drawings may include a front view, left and right side views, a rear view, a perspective view, and the like. Further, when the application is a trademark registration application, the plurality of processing target data may include at least one of the application form, the designated product, the designated service, and the trademark described in the application form. Here, the trademark may include characters, figures, colors only, sounds, holograms, and the like.

  Similar to the classification system 200 according to the first embodiment, the classification system 200 according to the present embodiment receives information on a plurality of processing target data according to the result output from the learning model in response to the input of the plurality of processing target data. A providing unit (output result providing unit 205) provided to the terminal 261, a receiving unit (teacher information receiving unit 207) that receives teacher information that is the source of teacher data from the user terminal 261, and teacher data corresponding to the teacher information And a teacher data input unit 203 for inputting to the model construction unit 201.

  Furthermore, the providing unit of the classification system 200 according to the present embodiment provides a processing target that is a basis for determining that at least one classification determined to be given to the application among a plurality of processing target data is given. Data is provided to the user terminal 261.

  FIG. 13 is a diagram illustrating another example of the UI screen on the user terminal. The UI screen shown in the figure is displayed on the user terminal 261, so that the user can select a plurality of input data related to the application, a plurality of classifications that can be assigned to the application provided from the classification system 200, and the basis for the assignment And can be confirmed.

  The user inputs data related to the application in the item of the input data 301. For example, when the application is a patent application, data relating to the application, the specification, the claims, the drawings, and the abstract may be input in the input data 301 item. When data is input to the items of the input data 301, items included in the input data 301 are listed on the item designation screen 302. The user designates the summary described in the abstract or all or a part of claims in the summary on the item designation screen 302, and a part of a plurality of processing target data to be input to the learning model Can be specified. Similarly, the paragraph specification screen 303 lists the paragraphs of the specification included in the input data 301. The user can specify a paragraph of the specification and specify some of a plurality of processing target data to be input to the learning model on the paragraph specifying screen 303. When the user finishes specifying the processing target data, the user presses the candidate extraction button 304, inputs the specified plurality of processing target data to the learning model, and outputs output data indicating at least one classification given to the application. obtain. The output data includes a value indicating the degree of approximation of each of the plurality of classifications that can be assigned to the application corresponding to the input data.

  The UI screen category candidates 305 indicate a plurality of categories that can be assigned to the application and the degree of approximation of each of the plurality of categories. In this example, a plurality of classifications that can be given to the application are IPCs “G06F 15/60” and “G06F 17/00”, and the degree of approximation is “approximate value 1”, “G06F 15/60” “Approximate value 0.8” for “G06F 17/00”. In this example, a plurality of classifications that can be assigned to the application are displayed by sorting in descending order of the degree of approximation, but the display order is arbitrary.

  When one of the plurality of classifications displayed in the classification candidate 305 is selected by the user, the item that is the basis for the assignment is displayed in the item-by-item grant basis description location 306, and The paragraph in the specification is displayed in the provision basis description portion 307 for each paragraph. Further, when one of the items displayed in the item-by-item grant basis description location 306 or the paragraph displayed in the paragraph-by-paragraph grant basis description location 307 is selected, the item or paragraph is displayed in the grant basis description location 308. Full text is displayed. Note that by double-clicking any one of the plurality of classifications displayed in the classification candidate 305, the description of the classification can be displayed. In this way, the user can confirm the validity of the classification by confirming the description that is the basis for the classification automatically given by the learning model.

  In the classification system 200 according to the present embodiment, when a plurality of items or a plurality of paragraphs are separately input to the learning model, items or paragraphs whose classification approximation indicated by the output data of the learning model is equal to or greater than a predetermined value. It may be determined as the grant basis. In addition, the classification system 200 inputs, when a plurality of items and a plurality of paragraphs are input to the learning model, and input data obtained by deleting a specific item or a specific paragraph from the plurality of items and the plurality of paragraphs, to the learning model. In some cases, the basis for assigning the classification may be determined based on how the output data changes. For example, when a specific item or a specific paragraph is deleted and the degree of approximation falls by a threshold value or more, the specific item or the specific paragraph may be used as the grant basis. Here, an example in which the granularity of the grant basis is a paragraph has been described, but a sentence unit, a phrase unit, and a word unit can be set as the grant basis in the same processing. And the structure which makes it highlight about the location used as the provision grounds may be sufficient. Examples of highlights include paragraphs, sentences, phrases, and words for the basis of grant, changing the background color of characters compared to other locations, changing the color of characters compared to other locations, and underlining characters For example, it may be drawn, bolded, or highlighted with a large font size. In addition, as an application example of FIG. 13, a function that further highlights a sentence, a phrase, or a word of the grant basis may be provided while displaying the paragraph that becomes the grant basis in the grant basis description location 308.

  The user selects a classification that is considered to best represent the characteristics of the application from among the plurality of classifications displayed in the classification candidate 305, and presses the first determination button 309, thereby selecting the first classification. Can be given to the application as In addition, the user selects a category that is considered to well represent the characteristics of the application from the plurality of categories displayed in the category candidate 305, and presses the classification determination button 310 to apply the category. Can be granted.

  When the processing target data that is the basis for determining that a certain classification is given to the application is included in the predetermined processing target data among the plurality of processing target data, the providing unit sets the classification as the first classification to the user terminal. May be provided. Here, the predetermined processing target data is data that is considered to well represent the characteristics of the application. For example, when the application is a patent application, a summary or a high-order claim that is considered to represent the characteristics of the invention well. It is description of a term.

  Further, the providing unit may provide the user terminal with the highest degree of approximation among the plurality of classifications that can be given to the application as the first classification. In addition, the classification system 200 may use the classification given based on the grant basis as the top classification when the description provided as the grant basis is a description that well represents the characteristics of the application.

  The processing target data input unit 204 of the classification system 200 inputs text data and image data related to the application as a plurality of input data to the learning model, and the model construction unit 201 is assigned to the application with the text data and image data. A learning model may be constructed based on teacher data indicating a relationship with output data indicating at least one classification. Here, when the application is a patent application, the text data may include an application, a specification, a claim, and an abstract, and the image data includes a chemical formula, a mathematical formula and a table inserted in the drawing and the specification. May include. In this case, the learning model is a single model, and the text data and the image data are learned as a set of data. The text data to be input may be limited to a paragraph in the specification describing the drawing, or may be obtained by performing OCR (Optical Character Recognition) processing on the drawing.

  The processing target data input unit 204 of the classification system 200 inputs text data related to the application as a part of the plurality of input data to the first sub-learning model included in the learning model, and inputs image data related to the application to the plurality of inputs. As part of the data, it is input to the second sub-learning model included in the learning model, and the model construction unit 201 indicates the teacher data indicating the relationship between the text data and the output data indicating at least one classification given to the application A first sub-learning model is constructed, and a second sub-learning model is constructed based on the teacher data indicating the relationship between the image data and the output data indicating at least one classification given to the application. Also good. In this case, the learning model includes a first sub learning model and a second sub learning model. Text data is input to the first sub learning model, and image data is input to the second sub learning model. Thus, it is good also as constructing | assembling the output data which shows the classification | category given to an application by constructing | assembling a sub learning model separately for every kind of data, and integrating the output data of a sub learning model.

  Classification by domestic and foreign patent offices and international organizations may be revised, including consolidation. Even in such a case, the classification system 200 according to the present embodiment may be used. When the classification is revised, the learning model is reconstructed based on the new classification, or the learning model is left constructed based on the old classification, and the output data of the learning model is converted to the old and new classification. That's good.

  In this example, an example in which the IPC is displayed in the classification candidate 305 has been described. However, the classification candidate 305 may display domestic and foreign patent classifications such as CPC, ECLA, ICO, FI, or F-term.

  When the input data is data related to a design registration application, the classification candidate 305 may display domestic and foreign design classifications such as design classification and D-term. Similarly, when the input data is data related to a trademark registration application, classification candidates 305 may display classifications related to domestic and foreign trademarks such as graphic classification of graphic trademarks, similar group codes of designated product services, and nice classification. . In this case, the learning model may include a trademark as input data, a graphic classification or the like as output data, a specified product service as input data, and a similar group code or a nice classification as output data. There may be.

  Note that the learning model may be constructed by teacher data using a designated product service as input data and an appropriate designated product service as output data. Here, the appropriate designated commodity service means a designated designated commodity service registered. In addition, the learning model may be constructed by teacher data using the trademark as input data and the designated product service as output data. As a result, a learning model for proposing a designated product service to be designated when applying for a similar trademark can be obtained. In addition, the learning model may be constructed by teacher data using a trademark as input data and a title as output data.

  The matters described in the fourth embodiment can be appropriately applied to each of the first, second, and third embodiments. For example, as an example 1, in addition to the in-house classification in the first, second, or third embodiment, It is set as the structure which provides the gazette classification of this Embodiment 4, and an in-house classification and a gazette classification can be provided about an in-house application. The publication classification assigned here can be used as the publication classification assigned by the applicant at the time of filing, and when the publication classification assigned by the JPO at the time of publication of the application is published and taken in as data, the embodiment 4 The granted publication classification may be overwritten on the publication classification given by the JPO. Further, as an example 2, as in the fourth embodiment, the basis for providing the publication classification is displayed, and in each of the first, second, and third embodiments, the basis for providing the in-house classification may be displayed.

  Conversely, the matters described in the first, second, or third embodiment can be applied to the fourth embodiment as appropriate. For example, the teacher information received from the multiple user terminals described in the first embodiment is integrated. Function (this function has a sub-function that is determined by majority when teacher information from a plurality of user terminals is different, a sub-function that determines teacher information in consideration of the weight for each user, and teacher information based on instructions from the reader It is also possible to add a sub-function for determining the above) to the fourth embodiment.

  Each unit for realizing the functions of the above-described embodiments can be implemented by, for example, hardware or software. When implemented by software, program code for controlling hardware may be executed by various processors such as a CPU and MPU. Hardware such as a circuit for realizing the function of the program code may be provided. A part of the program code may be realized by hardware, and the remaining part may be executed by various processors.

DESCRIPTION OF SYMBOLS 200 Classification system 201 Model construction part 202 Learning model 203 Teacher data input part 204 Process target data input part 205 Output result provision part 205
206 Destination DB
207 Teacher information reception unit 208 Teacher information integration unit 209 Teacher information reflection unit 210 Classified DB
250 Gazette DB
261 User terminal

Claims (30)

Model construction for constructing a learning model based on teacher data indicating a relationship between input data relating to at least one publication and output data indicating at least one classification assigned to at least one publication corresponding to the input data And
Processing target data input unit for inputting processing target data relating to at least one publication as the input data to the learning model;
A providing unit that provides information on the processing target data to a user terminal according to a result output from the learning model in response to an input of the processing target data;
A receiving unit that receives teacher information as a source of the teacher data from the user terminal;
A classification system comprising: a teacher data input unit that inputs the teacher data corresponding to the teacher information to the model construction unit. The teacher information transmitted from the user terminal includes information indicating that a specific classification of a specific publication is not appropriate,
The classification system according to claim 1, wherein the model construction unit constructs a learning model that does not assign the specific classification to the specific publication based on the teacher data corresponding to the teacher information.   The classification according to claim 1, wherein the output data indicated by the teacher data includes a value indicating a degree of approximation of each of a plurality of classifications that can be assigned to at least one of the publications corresponding to the input data. system.   The classification system according to claim 3, wherein the value indicating the degree of approximation includes a value indicating that the approximation is performed or a value indicating that the approximation is not performed.   The classification system according to claim 4, wherein the value indicating the degree of approximation further includes an intermediate value between a value indicating approximation and a value indicating not approximation. The receiving unit receives teacher information including an evaluation,
The classification system according to claim 3, wherein the value indicating the degree of approximation is determined according to the evaluation. The processing target data input unit inputs processing target data related to a plurality of publications to the learning model,
The said provision part provides the list | wrist of the gazette which changed the arrangement | sequence order of several gazettes in the order with the high value which shows that it is approximated to the said user terminal. Classification system.   The classification system according to claim 7, wherein the plurality of publications are newly arrived publications issued in a predetermined period. The providing unit includes:
Get information to identify users corresponding to each classification,
The information on the processing target data is provided to the user terminal of the user corresponding to the classification output from the learning model based on the acquired information specifying the user. Classification system according to paragraph. The providing unit includes:
When the approximation output from the learning model in response to the input of the processing target data exceeds a predetermined threshold, the processing target data is transferred to the user terminal of the user corresponding to the classification having the approximation exceeding the threshold. Offer to,
10. The classification system according to claim 9, wherein when the degree of approximation is equal to or less than a predetermined threshold, the processing target data is not provided to the user terminal of the user corresponding to a classification having an degree of approximation equal to or less than the threshold. When the degree of approximation output from the learning model in response to an input of the processing target data exceeds a threshold regarding a plurality of classifications, the providing unit sends the processing target data to the user terminal of the user corresponding to each classification Provide information about
The receiving unit receives teacher information from each user terminal,
The classification system according to claim 9 or 10, further comprising an integration unit that integrates teacher information received from a plurality of user terminals.   The classification system according to claim 11, wherein when the teacher information of some user terminals is different from the teacher information of other user terminals, the integration unit determines the teacher information to be adopted by majority vote. The integration unit
Get weight information set for each user,
The classification system according to claim 11, wherein when the teacher information of some user terminals is different from the teacher information of other user terminals, the teacher information is determined in consideration of a weight for each user. The integration unit
The reader can be identified among users,
12. The classification system according to claim 11, wherein when the teacher information of some user terminals is different from the teacher information of other user terminals, an instruction from the reader is received and the teacher information is determined based on the received instruction. . Input data relating to at least one publication, output data indicating at least one classification assigned to at least one publication corresponding to the input data, and output indicating a degree of risk assigned to the publication corresponding to the input data A model building unit that builds a learning model based on teacher data indicating a relationship with the data;
Processing target data input unit for inputting processing target data relating to at least one publication as the input data to the learning model;
A providing unit that provides information on the processing target data to a user terminal according to a result output from the learning model in response to an input of the processing target data;
A receiving unit that receives teacher information as a source of the teacher data from the user terminal;
A classification system comprising: a teacher data input unit that inputs the teacher data corresponding to the teacher information to the model construction unit. A model construction unit that constructs a learning model based on teacher data indicating a relationship between input data relating to at least one publication and output data indicating a degree of risk assigned to the publication corresponding to the input data;
Processing target data input unit for inputting processing target data relating to at least one publication as the input data to the learning model;
A providing unit that provides information on the processing target data to a user terminal according to a result output from the learning model in response to an input of the processing target data;
A receiving unit that receives teacher information as a source of the teacher data from the user terminal;
A classification system comprising: a teacher data input unit that inputs the teacher data corresponding to the teacher information to the model construction unit. Model construction for constructing a learning model based on teacher data indicating a relationship between input data relating to at least one publication and output data indicating at least one classification assigned to at least one publication corresponding to the input data A method for controlling a classification system having parts,
Processing target data input step for inputting processing target data relating to at least one publication as the input data to the learning model;
Providing a user terminal with information on the processing target data according to a result output from the learning model in response to an input of the processing target data;
A receiving step of receiving teacher information as a source of the teacher data from the user terminal;
A method for controlling a classification system, comprising: a teacher data input step of inputting the teacher data corresponding to the teacher information into the model construction unit.   The program for functioning a computer as each part of the classification system as described in any one of Claims 1-16. A model construction unit that constructs a learning model based on teacher data indicating a relationship between a plurality of input data relating to an application and output data indicating at least one classification assigned to the application;
A plurality of processing target data related to the application, a processing target data input unit that inputs the plurality of processing data as the plurality of input data to the learning model;
A providing unit for providing information on the plurality of processing target data to a user terminal according to a result output from the learning model in response to an input of the plurality of processing target data;
A receiving unit that receives teacher information as a source of the teacher data from the user terminal;
A teacher data input unit that inputs the teacher data corresponding to the teacher information into the model construction unit;
The providing unit provides the user terminal with processing target data that is a basis for determining that the at least one classification is given to the application among the plurality of processing target data.
Classification system.   When the processing target data that is the basis for determining that the at least one classification is given to the application is included in predetermined processing target data among the plurality of processing target data, the providing unit determines the classification. The classification system according to claim 19, wherein the classification system is provided to the user terminal as a first classification. The output data indicated by the teacher data includes a value indicating the degree of approximation of each of a plurality of classifications that can be assigned to the application corresponding to the input data,
21. The classification system according to claim 19 or 20, wherein the providing unit provides the user terminal with the highest degree of approximation among a plurality of classifications that can be assigned to the application as a top classification. The processing target data input unit inputs text data and image data related to the application as the plurality of input data to the learning model,
The model construction unit constructs the learning model based on teacher data indicating a relationship between the text data and image data and output data indicating at least one classification given to the application. 22. The classification system according to any one of 21. The processing target data input unit inputs text data relating to the application as a part of the plurality of input data to a first sub-learning model included in the learning model, and image data relating to the application is input to the plurality of pieces of data. As part of the input data, input to the second sub-learning model included in the learning model,
The model construction unit constructs the first sub-learning model based on teacher data indicating a relationship between the text data and output data indicating at least one classification given to the application, and the image data The classification according to any one of claims 19 to 21, wherein the second sub-learning model is constructed based on teacher data indicating a relationship with output data indicating at least one classification assigned to the application. system. The application is a patent application or a utility model registration application,
The plurality of pieces of processing target data include at least one of an application, a specification, a claim, or a utility model registration claim, a drawing, and a summary, according to any one of claims 19 to 23. Classification system. The application is a design registration application,
The plurality of pieces of processing target data include at least one of an application, an article relating to the design described in the application, an explanation of the article relating to the design, an explanation of the design, and a drawing. Classification system according to paragraph. The application is a trademark registration application;
The classification system according to any one of claims 19 to 23, wherein the plurality of pieces of processing target data include at least one of an application and a specified product, a specified service, and a trademark described in the application. The output data indicated by the teacher data includes a value indicating the degree of approximation of each of a plurality of classifications that can be assigned to at least one of the applications corresponding to the input data,
The providing unit includes:
Get information to identify users corresponding to each classification,
Based on the acquired information identifying the user, providing information on the processing target data to the user terminal of the user corresponding to the classification output from the learning model,
When the degree of approximation output from the learning model in response to the input of the processing target data exceeds a threshold for a plurality of classifications, information on the processing target data is provided to the user terminal of the user corresponding to each classification. ,
The receiving unit receives teacher information from each user terminal,
The classification system according to any one of claims 19 to 26, further comprising an integration unit that integrates teacher information received from a plurality of user terminals.   The classification system according to claim 27, wherein when the teacher information of some user terminals is different from the teacher information of other user terminals, the integration unit determines the teacher information to be adopted by majority vote. The integration unit
Get weight information set for each user,
The classification system according to claim 27, wherein when the teacher information of some user terminals is different from the teacher information of other user terminals, the teacher information is determined in consideration of the weight for each user. The integration unit
The reader can be identified among users,
28. The classification system according to claim 27, wherein when the teacher information of some user terminals is different from the teacher information of other user terminals, an instruction from the reader is received and the teacher information is determined based on the received instruction. .

Images