Sentence generation model generation device, sentence generation model generation method, and sentence generation device

Abstract

A sentence generation model generation device (10) comprises a training unit (11) that, in response to the input of an input sentence, identifies, from the input sentence, an assertion and a ground for reinforcing the assertion from the same standpoint as or a different standpoint from the assertion, executes at least one of labeling of the assertion and the ground and labeling of the relationship between the assertion and the ground, sets the input sentence as an explanatory variable, and conducts machine learning in which the text at a point where the labeling was executed is used as an objective variable, thereby conducting training of a sentence generation model for generating and outputting a text sentence related to the structure of the input sentence and generating a sentence generation model.

Description

Sentence generation model generating device, sentence generation model generating method, and sentence generation device

[0001] The present disclosure relates to a sentence generation model generation device, a sentence generation model generation method, and a sentence generation device.

[0002] One example of a task for structuring a document is a technique called argument mining, which identifies the author's claims and evidence from the document and reveals their logical structure. For example, Non-Patent Document 1 below proposes a multi-task learning method that simultaneously performs three tasks: (1) extracting arguments (parts containing claims and evidence) from the document, (2) labeling components (components: claims and evidence), and (3) labeling relations (relationships: relationships between claims and evidence).

[0003] Gaku Morio et al.'s paper "End-to-end Argument Mining with Cross-corpora Multi-task Learning," published in May 2022

[0004] The technology described in Non-Patent Document 1 requires the above three tasks to be performed simultaneously, which requires complex parameter adjustments to achieve a certain level of accuracy in the document structuring task, making it difficult to simplify the processing and further improve performance.

[0005] In view of the above circumstances, the present disclosure aims to simplify processing and improve performance by eliminating the complex parameter adjustments that have traditionally been required to achieve a certain level of accuracy in document structuring tasks.

[0006] A sentence generation model generation device according to the present disclosure includes a learning unit that, in response to input of an input sentence, identifies from the input sentence an assertion and grounds that support the assertion from the same or different standpoint as the assertion, performs at least one of labeling the assertion and the grounds and labeling the relationship between the assertion and the grounds, and performs machine learning using the input sentence as an explanatory variable and the text of the labeled portion as an objective variable, thereby training a sentence generation model to generate and output text sentences related to the structure of the input sentence, thereby generating the sentence generation model. Note that the "assertion" corresponds to, for example, a claim in a discussion mining task or a claim in a structuring task targeting posts to a social networking service (SNS), and the "grounds" corresponds to, for example, a premises in a discussion mining task or evidence in a structuring task targeting posts to a social networking service (SNS).

[0007] According to the present disclosure, a sentence generation model whose input and output are text can be obtained by the above-described machine learning, and by using the trained sentence generation model for a document structuring task, the complicated parameter adjustment that was previously required to achieve a certain level of accuracy is no longer necessary, and processing can be simplified and performance can be improved. Furthermore, the present disclosure can be applied not only to discussion mining tasks but also to various tasks involving document structuring.

[0008] 1 is a configuration diagram of a sentence generation device configured to include a sentence generation model generation device according to an embodiment. FIG. 2 is a flow diagram showing processing in the learning phase of a sentence generation model. FIG. 3 is a diagram for explaining argument identification in an argument mining task, FIG. 4 is a diagram for explaining component labeling in an argument mining task, and FIG. 5 is a diagram for explaining relation labeling in an argument mining task. FIG. 4 is a diagram showing an example of a text sentence representing a tree structure. FIG. 5 is a flow diagram showing processing in the inference phase. FIG. 6 is a diagram showing a document including a major claim, a claim, and a premise. FIG. 7 is a diagram showing the document structure of the document shown in FIG. 6. FIG. 7 is a diagram showing an example of an input sentence, and FIG. 8 is a diagram showing the document structure of the input sentence in FIG. 8. FIG. 9 is a diagram for explaining the effect of an embodiment. FIG. 9 is a diagram showing an example of an input sentence, and FIG. 10 is a diagram showing an example of an output when a portion not including a component is output, and FIG. 11 is a diagram showing an example of an output when a portion not including a component is not output. FIG. 10 is a diagram for explaining the effect of not outputting a portion not including a component. FIG. 11 is a diagram showing an example of an input sentence, and FIG. 12 is a diagram showing an example of an output for pattern A (subtasks (1) to (3)). (a) is a diagram showing an example of output in the case of pattern B (only subtasks (1) and (2)), and (b) is a diagram showing an example of output in the case of pattern C (only subtasks (1) and (3)). (a) is a diagram showing a first modified example of the system configuration, (b) is a diagram showing a second modified example of the system configuration, and (c) is a diagram showing a third modified example of the system configuration. (b) is a diagram showing an example of the hardware configuration of a sentence generation model generation device.

[0009] An embodiment of the present disclosure will be described below with reference to the drawings. The technology of the present disclosure can be applied to structuring various documents, such as (1) structuring essays defined in three levels: major claims (major assertions), claims (normal assertions), and premises (evidence), (2) structuring essays defined in two levels: claims (assertions) and premises (evidence), and (3) structuring social media posts defined in two levels: claims (assertions) and evidence (evidence corresponding to the evidence). However, first, an example of applying the present disclosure to a discussion mining task that performs structuring of essays defined in two levels: claims and premises will be described.

[0010] An example of a system configuration is shown in Fig. 1. As shown in Fig. 1, the sentence generation device 1 includes a sentence generation model generation device 10 that generates a sentence generation model M, which is a learning model for generating and outputting a text sentence related to the structure of an input sentence. Furthermore, since both the input and output of the sentence generation model M are text sentences, as will be described later, the sentence generation model M is configured using a pre-learning model of a general encoder / decoder, thereby eliminating the need for parameter adjustment as in the past.

[0011] The sentence generation model generation device 10 includes, as functional blocks for realizing the functions of the present disclosure, a learning unit 11 that performs learning of a sentence generation model M to generate a sentence generation model M, and a model holding unit 12 that holds the generated sentence generation model M. The sentence generation device 1 includes the sentence generation model generation device 10 described above, and a text sentence acquisition unit 13 that inputs a target input sentence into the sentence generation model M and acquires a text sentence related to the structure of the target input sentence as output from the sentence generation model M. Details of the functions of each unit will be described below along with the explanation of the processing based on Figures 2 to 5. Note that the system configuration is not limited to the configuration of Figure 1, and other system configurations are also possible. Other system configurations will be described later using Figures 13(a) to 13(c).

[0012] (Processing Executed in the Sentence Generation Model Generating Device 10: Processing in the Learning Phase) Next, the processing (processing in the learning phase) executed by the sentence generation model generating device 10 will be described with reference to the flow diagram of FIG.

[0013] First, the learning unit 11 in the sentence generation model generation device 10 receives an externally input sentence (step S1 in FIG. 2 ) and identifies arguments (portions containing a claim and a premise) from the argument mining task (step S2). Here, "claim" corresponds to an assertion, and "premises" corresponds to evidence supporting the claim from the same or a different standpoint. For example, the learning unit 11 identifies the underlined portions of the input sentence shown in FIG. 3( a), i.e., "society should be educated and became aware of health issues" and "the potential precautions on the way of illnesses can be taken instead of trying to provide treatment for the increasing number of patients," as arguments.

[0014] Next, the learning unit 11 labels the components (claims and premises) (step S3). For example, as shown in FIG. 3B, the learning unit 11 labels "society should be educated and became aware of health issues" identified in step S2 as a claim, and labels "the potential precautions on the way of illnesses can be taken instead of trying to provide treatment for the increasing number of patients" as a premises.

[0015] Next, the learning unit 11 labels the relationship between the claim and the premises (step S4). In this case, since the premises reinforce the claim from the same perspective as the claim, the learning unit 11 labels the relationship between the claim and the premises as "support" as shown in Figure 3(c). On the other hand, if the premises reinforce the claim from the opposite perspective, the learning unit 11 labels the relationship between the claim and the premises as "attack."

[0016] The learning unit 11 then performs machine learning using the input sentence as an explanatory variable and the text sentence for the target portion of the labeling as a target variable (step S5). The text sentence used as the target variable here is, for example, as shown in Fig. 4, where the part of the input sentence (Fig. 3(a)) labeled as a claim is expressed as [society should be educated and became aware of health issues | claim for], and the part labeled as a premise is expressed as [the potential precautions on the way of illnesses can be taken instead of trying to provide treatment for the increasing number of patients | premise | support = society should be educated and became aware of health issues]. Here, the latter part of the part labeled as premise, "support = society should be educated and became aware of health issues," represents a relation that supports the claim (reinforces it from the same perspective).

[0017] The above processing of steps S1 to S5 corresponds to learning of the sentence generation model. Furthermore, the model holding unit 12 holds the trained sentence generation model M (step S6). In this way, the trained sentence generation model M is generated and held.

[0018] (Processing Executed in Sentence Generation Device 1: Processing in Inference Phase) Next, the processing executed by the sentence generation device 1 (processing in the inference phase) will be described with reference to the flow diagram of FIG.

[0019] The text sentence acquisition unit 13 in the sentence generation device 1 receives an input sentence input from outside (step S11 in FIG. 5), and inputs the input sentence to the sentence generation model M, thereby acquiring a text sentence representing the tree structure of the input sentence as an output from the sentence generation model M (step S12). The "text sentence representing the tree structure" acquired here has the claim and premises each shown in quotation marks, as in the example of FIG. 5 described above, and furthermore, the latter half of the premises represents the relationship between the claim and premises described above (in the example of FIG. 5, "support" means that the premises supports the claim from the same standpoint).

[0020] In the embodiment described above, a sentence generation model in which both input and output are text can be generated by the process in Figure 2. By using the obtained trained sentence generation model for a document structuring task, the complicated parameter adjustment that was previously required to achieve a certain level of accuracy is no longer necessary, and processing can be simplified and performance can be improved. Furthermore, the present invention can be applied not only to discussion mining tasks but also to various tasks involving document structuring.

[0021] (Another aspect) Here, we will explain the structuring of essays defined in three stages, major claim (major assertion), claim (normal assertion), and premise (ground), as mentioned at the beginning of the embodiment, using Figures 6 to 8.

[0022] Similar to the relationship between a "claim" and a "premise" described above, the relationship between a "major claim" and a "claim" can also be understood. That is, in one document, one or more "claims" that reinforce a major claim from the same or opposite position as the "major claim" can be understood, and a "for" relationship is defined when a claim reinforces a major claim from the same position as the "major claim," and an "against" relationship is defined when a claim reinforces a major claim from the opposite position to the "major claim."

[0023] Figure 6 shows an example of a document in which arguments (including major claims, claims, and premises) have been identified in the argument mining task, and Figure 7 shows an example of a document structure in which the two major claims, two claims, and six premises identified in Figure 6 have been labeled, and their relationships have also been labeled.

[0024] As a specific example, when the input sentence shown in FIG. 8(a) is input, one major claim, two claims, and two premises are labeled as shown in FIG. 8(b), and the relationships between the major claim and the claims (both of which are "for") and the relationships between the claims and the premises (both of which are "support") are labeled.

[0025] FIG. 9 is a table showing the "match rate (unit: %)," which is the performance of each of the existing models and the model to which the present disclosure is applied. The higher the match rate, the better the performance. Note that C in the table represents the match rate of labeled components, and R in the table represents the match rate of labeled relations. The match rates are shown as evaluation values ​​for "document unit" and "paragraph unit." Among the models to which the present disclosure is applied in FIG. 9 , "T5 base" is a natural language processing model (model size: 220 MB) published by Google (registered trademark), and "Flan-T5 base" is a T5 base with improved zero-shot learning. Similarly, "T5 large" is a natural language processing model (model size: 770 MB) published by Google, and "Flan-T5 large" is a T5 large with improved zero-shot learning. The underlined evaluation values ​​(match rates) in the table in FIG. 9 indicate that the model to which the present disclosure is applied has better performance than any existing model, i.e., that the present disclosure improves inference performance.

[0026] (Improving Points Regarding Output Format) The following describes the improvements regarding the output format in the present disclosure. Specifically, the learning unit 11 performs learning using text containing only identified components (claims and evidence) as the objective variable in machine learning. In other words, learning is performed so that the text sentence output from the sentence generation model M becomes text that does not contain any parts other than the components (claims and evidence) (i.e., text containing only the components). In practice, it is sufficient to output only the identified components, and there is little need to output the entire document including extra parts (parts other than the identified components). On the other hand, if the extra parts are output, there is a concern that the number of decoding operations during processing in the pre-training model of the encoder / decoder will increase, increasing the decoding processing time and ultimately resulting in a decrease in performance. Therefore, the above-mentioned improvements can alleviate the above concerns and improve performance.

[0027] For example, when outputting a "portion that does not contain components (assertion and evidence)" in the input sentence shown in Fig. 10(a), even the "portion that does not contain components," "I completely agree with the idea that," is output as shown in Fig. 10(b). On the other hand, by setting the objective variable in machine learning to text that contains only the identified components, the learning unit 11 generates a sentence generation model M that does not output a "portion that does not contain components," "I completely agree with the idea that," as shown in Fig. 10(c).

[0028] FIG. 11 shows the performance (match rate (unit: %)) before and after changing the output format for each of the aforementioned "Flan-T5 base" and "Flan-T5 large." Note that C in the table represents the match rate of labeled components, and R in the table represents the match rate of labeled relations.

[0029] From Figure 11, it can be seen that for both the "Flan-T5 base" and "Flan-T5 large" models, the matching rate of labeled components and the matching rate of labeled relations are improved after the output format change compared to before (i.e., performance is improved).

[0030] (Regarding a Modified Example of the Processing in FIG. 2) In the above-described embodiment, an example was described in which the present disclosure was applied to an argument mining task that performs all three subtasks: (1) identifying arguments, (2) labeling components, and (3) labeling relations. However, in practice, there are cases in which it is not necessary to perform all of the subtasks (1) to (3). For example, if you want to label only the claims and premises contained in an input sentence and determine their total number, or if you want to determine the locations of claims and premises in the input sentence, you only need to perform the subtasks (1) and (2). Also, if you only want to determine whether a certain structure exists in the input sentence, regardless of the labels such as claims and premises, you only need to perform the subtasks (1) and (3).

[0031] As a specific example, if the input sentence shown in FIG. 12(a) is input, the output example of pattern A (when all subtasks (1) to (3) are performed) will be the text sentence shown in FIG. 12(b), which is a text sentence representing a tree structure including labeled components and labeled relations.

[0032] On the other hand, an output example of pattern B (when only subtasks (1) and (2) are performed) is the text sentence shown in Figure 13(a). Although it does not include labeled relations, the labeled components (major claim, claim, premise) are shown as underlined parts, so that the total number of each component can be grasped, and also the location of each component in the input sentence can be grasped. Note that in Figure 13(a), labeled components are underlined for the sake of convenience, but the actual output text sentence is not underlined. The same applies to Figure 13(b).

[0033] Furthermore, an example output from Pattern C (when only subtasks (1) and (3) are performed) is the text sentence shown in Figure 13(b). Although it does not contain labeled components, labeled relations (here, three "support" relations) are shown as underlined parts, so it can be seen that some structure exists in the input sentence.

[0034] (Regarding Modified System Configurations) As described above, the system configuration is not limited to the configuration of FIG. 1, and other system configurations exemplified in FIGS. 14(a) to 14(c) may be adopted. For example, as shown in FIG. 14(a), the sentence generation model generation device 10, the model holding unit 12, and the sentence generation device 1 may be configured as devices independent of each other. Also, as shown in FIG. 14(b), the sentence generation model generation device 10 may be configured as a device independent of the sentence generation device 1, and the model holding unit 12 may be included in the sentence generation model generation device 10. Furthermore, as shown in FIG. 14(c), the sentence generation model generation device 10 may be configured as a device independent of the sentence generation device 1, and the model holding unit 12 may be included in the sentence generation device 1. Even with the various system configurations described above, the same effects as those of the above-described embodiment can be achieved.

[0035] The gist of the present disclosure lies in the following [1] to [5]. [1] A sentence generation model generation device comprising: a learning unit that, in response to an input of an input sentence, identifies from the input sentence an assertion and grounds that support the assertion from the same or a different standpoint as the assertion, performs at least one of labeling the assertion and the grounds and labeling the relationship between the assertion and the grounds, and performs machine learning using the input sentence as an explanatory variable and a text at a location where the labeling has been performed as a target variable, thereby training a sentence generation model to generate and output a text sentence related to the structure of the input sentence, thereby generating the sentence generation model. [2] The sentence generation model generation device described in [1], wherein the learning unit further identifies, as the assertions, a major assertion that is a main assertion and a normal assertion that supports the major assertion from the same or a different standpoint as the major assertion, and performs at least one of labeling the major assertion and the normal assertion and labeling the relationship between the major assertion and the normal assertion. [3] The sentence generation model generation device according to [1] or [2], wherein the learning unit sets the objective variable in the machine learning to a text including only the identified claim and the evidence. [4] A sentence generation model generation method executed by a sentence generation model generation device, comprising: a step of generating the sentence generation model by conducting learning of a sentence generation model for generating and outputting text sentences related to the structure of an input sentence, the step of generating the sentence generation model including: a substep of identifying, from the input sentence, a claim and evidence that supports the claim from the same standpoint as the claim or a standpoint different from the claim; a substep of performing at least one of labeling the claim and the evidence and labeling the relationship between the claim and the evidence; and a substep of performing machine learning using the input sentence as an explanatory variable and text at a portion where the labeling has been performed as an objective variable.[5] A sentence generation device comprising: a text sentence acquisition unit that acquires a text sentence relating to the structure of a target input sentence as an output from a sentence generation model generated by the sentence generation model generation device according to any one of [1] to [3] above, by inputting a target input sentence into the sentence generation model.

[0036] (Explanation of terms, explanation of hardware configuration (Figure 15), etc.) Note that the block diagrams used in the explanation of the above embodiments show functional blocks. These functional blocks (components) are realized by any combination of at least one of hardware and software. Furthermore, the method of realizing each functional block is not particularly limited. That is, each functional block may be realized using one device that is physically or logically coupled, or may be realized using two or more devices that are physically or logically separated and connected directly or indirectly (for example, using wires, wirelessly, etc.) and these multiple devices. A functional block may be realized by combining software with the one device or the multiple devices.

[0037] Functions include, but are not limited to, judgment, determination, judgment, calculation, computation, processing, derivation, investigation, search, confirmation, reception, transmission, output, access, resolution, selection, selection, establishment, comparison, assumption, expectation, consideration, broadcasting, notifying, communicating, forwarding, configuring, reconfiguring, allocating, mapping, and assignment. For example, a functional block (component) that performs transmission is called a transmitting unit or transmitter. As mentioned above, there are no particular limitations on how these functions are implemented.

[0038] For example, a sentence generation model generation device according to an embodiment of the present disclosure may function as a computer that executes the processes of the present disclosure. Fig. 15 is a diagram illustrating an example of the hardware configuration of a sentence generation model generation device 10 according to an embodiment of the present disclosure. The above-described sentence generation model generation device 10 may be physically configured as a computer device including a processor 1001, a memory 1002, a storage 1003, a communication device 1004, an input device 1005, an output device 1006, a bus 1007, and the like. The hardware configuration of the sentence generation device 1 may also be the same as that of the sentence generation model generation device 10.

[0039] In the following description, the term "apparatus" can be interpreted as a circuit, a device, a unit, etc. The hardware configuration of the sentence generation model generation device 10 may be configured to include one or more of the devices shown in the figure, or may be configured to exclude some of the devices.

[0040] Each function of the sentence generation model generating device 10 is realized by loading specific software (programs) onto hardware such as the processor 1001 and memory 1002, causing the processor 1001 to perform calculations, control communication via the communication device 1004, and control at least one of reading and writing data from and to the memory 1002 and storage 1003.

[0041] The processor 1001 controls the entire computer by running, for example, an operating system, and may be configured as a central processing unit (CPU) including an interface with peripheral devices, a control device, an arithmetic unit, a register, etc.

[0042] The processor 1001 also reads programs (program codes), software modules, data, etc. from at least one of the storage 1003 and the communication device 1004 into the memory 1002 and executes various processes in accordance with these. The programs used are those that cause a computer to execute at least some of the operations described in the above-described embodiments. While the various processes have been described as being executed by one processor 1001, they may be executed simultaneously or sequentially by two or more processors 1001. The processor 1001 may be implemented by one or more chips. The programs may be transmitted from a network via a telecommunications line.

[0043] The memory 1002 is a computer-readable recording medium and may be configured by, for example, at least one of a read-only memory (ROM), an erasable programmable ROM (EPROM), an electrically erasable programmable ROM (EEPROM), a random access memory (RAM), etc. The memory 1002 may also be called a register, a cache, a main memory (primary storage device), etc. The memory 1002 can store executable programs (program codes), software modules, etc. for implementing a wireless communication method according to an embodiment of the present disclosure.

[0044] Storage 1003 is a computer-readable recording medium, and may be composed of at least one of, for example, an optical disk such as a CD-ROM (Compact Disc ROM), a hard disk drive, a flexible disk, a magneto-optical disk (e.g., a compact disk, a digital versatile disk, a Blu-ray (registered trademark) disk), a smart card, a flash memory (e.g., a card, a stick, a key drive), a floppy (registered trademark) disk, a magnetic strip, etc. Storage 1003 may also be referred to as an auxiliary storage device. The above-mentioned storage medium may be, for example, a database, a server, or other appropriate medium including at least one of memory 1002 and storage 1003.

[0045] The communication device 1004 is hardware (transmission / reception device) for communicating between computers via at least one of a wired network and a wireless network, and is also referred to as, for example, a network device, a network controller, a network card, a communication module, etc. The communication device 1004 may be configured to include a high-frequency switch, a duplexer, a filter, a frequency synthesizer, etc. to realize at least one of frequency division duplex (FDD) and time division duplex (TDD).

[0046] The input device 1005 is an input device (e.g., a keyboard, a mouse, a microphone, a switch, a button, a sensor, etc.) that receives input from the outside. The output device 1006 is an output device (e.g., a display, a speaker, an LED lamp, etc.) that outputs to the outside. The input device 1005 and the output device 1006 may be integrated into one device (e.g., a touch panel).

[0047] Furthermore, each device, such as the processor 1001 and the memory 1002, is connected by a bus 1007 for communicating information. The bus 1007 may be configured using a single bus, or may be configured using different buses between each device.

[0048] The sentence generation model generation device 10 may also be configured to include hardware such as a microprocessor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a programmable logic device (PLD), or a field programmable gate array (FPGA), and some or all of the functional blocks may be realized by the hardware. For example, the processor 1001 may be implemented using at least one of these pieces of hardware.

[0049] The notification of information is not limited to the aspects / embodiments described in the present disclosure and may be performed using other methods. For example, the notification of information may be performed by physical layer signaling (e.g., Downlink Control Information (DCI) and Uplink Control Information (UCI)), higher layer signaling (e.g., Radio Resource Control (RRC) signaling, Medium Access Control (MAC) signaling, broadcast information (Master Information Block (MIB) and System Information Block (SIB))), other signals, or a combination thereof. Furthermore, the RRC signaling may be referred to as an RRC message, and may be, for example, an RRC Connection Setup message, an RRC Connection Reconfiguration message, or the like.

[0050] Each aspect / embodiment described in the present disclosure may be implemented using any of the following standards: LTE (Long Term Evolution), LTE-Advanced (LTE-A), SUPER 3G, IMT-Advanced, 4G (4th generation mobile communication system), 5G (5th generation mobile communication system), 6th generation mobile communication system (6G), xth generation mobile communication system (xG) (xG (x is, for example, an integer or a decimal number)), FRA (Future Radio Access), NR (new Radio), New radio access (NX), Future generation radio access (FX), W-CDMA (registered trademark), GSM (registered trademark), CDMA2000, UMB (Ultra Mobile Broadband), IEEE 802.11 (Wi-Fi (registered trademark)), IEEE 802.16 (WiMAX (registered trademark)), IEEE 802.17 (WiMAX (registered trademark)), IEEE 802.19 (WiMAX (registered trademark)), IEEE 802.20 (WiMAX (registered trademark)), IEEE 802.21 (Wi-Fi (registered trademark)), IEEE 802.22 (WiMAX (registered trademark)), IEEE 802.23 (WiMAX (registered trademark)), IEEE 802.24 (WiMAX (registered trademark)), IEEE 802.25 (WiMAX (registered trademark)), IEEE 802.26 (WiMAX (registered trademark)), IEEE 802.27 (WiMAX (registered trademark)), IEEE 802.28 (WiMAX (registered trademark)), IEEE 802.29 (WiMAX (registered trademark)), IEEE 802.30 (WiMAX (registered trademark)), IEEE 802.31 (Wi-Fi (registered trademark)), IEEE 802.32 (WiMAX (registered trademark)), IEEE 802.33 (WiMAX (registered trademark)), IEEE 802.34 ( The present invention may be applied to at least one of systems using 802.20, UWB (Ultra-Wide Band), Bluetooth (registered trademark), or other suitable systems, and next-generation systems that are extended, modified, created, or defined based on these systems. The present invention may also be applied to a combination of multiple systems (e.g., a combination of LTE and / or LTE-A with 5G).

[0051] The order of the procedures, sequences, flowcharts, etc. of each aspect / embodiment described in this disclosure may be changed unless it is consistent. For example, the methods described in this disclosure present elements of various steps using an example order, and are not limited to the particular order presented.

[0052] Input and output information may be stored in a specific location (for example, memory) or may be managed using a management table. Input and output information may be overwritten, updated, or added to. Output information may be deleted. Input information may be transmitted to another device.

[0053] The determination may be made based on a value represented by one bit (0 or 1), a Boolean value (true or false), or a numerical comparison (e.g., comparison with a predetermined value).

[0054] The aspects / embodiments described in this disclosure may be used alone, in combination, or switched depending on the implementation. Notification of predetermined information (e.g., notification that "X is true") is not limited to explicit notification, but may be implicit (e.g., not notifying the predetermined information).

[0055] Although the present disclosure has been described in detail above, it is clear to those skilled in the art that the present disclosure is not limited to the embodiments described herein. The present disclosure can be implemented in modified and altered forms without departing from the spirit and scope of the present disclosure as defined by the claims. Therefore, the description of the present disclosure is intended to be illustrative and does not have any limiting meaning on the present disclosure.

[0056] Software shall be construed broadly to mean instructions, instruction sets, code, code segments, program code, programs, subprograms, software modules, applications, software applications, software packages, routines, subroutines, objects, executable files, threads of execution, procedures, functions, etc., whether referred to as software, firmware, middleware, microcode, hardware description language, or otherwise.

[0057] Software, instructions, information, etc. may also be transmitted or received over a transmission medium. For example, if software is transmitted from a website, server, or other remote source using wired technologies (such as coaxial cable, fiber optic cable, twisted pair, Digital Subscriber Line (DSL)), and / or wireless technologies (such as infrared, microwave), then these wired and / or wireless technologies are included within the definition of transmission media.

[0058] The information, signals, etc. described in this disclosure may be represented using any of a variety of different technologies. For example, data, instructions, commands, information, signals, bits, symbols, chips, etc. that may be referred to throughout the above description may be represented by voltages, currents, electromagnetic waves, magnetic fields or magnetic particles, optical fields or photons, or any combination thereof.

[0059] Note that terms described in this disclosure and terms necessary for understanding this disclosure may be replaced with terms having the same or similar meanings. For example, at least one of a channel and a symbol may be a signal (signaling). Furthermore, a signal may be a message. Furthermore, a component carrier (CC) may be called a carrier frequency, a cell, a frequency carrier, etc.

[0060] As used in this disclosure, the terms "system" and "network" are used interchangeably.

[0061] Furthermore, the information, parameters, etc. described in the present disclosure may be expressed using absolute values, relative values ​​from a predetermined value, or other corresponding information. For example, a radio resource may be indicated by an index.

[0062] The names used for the above-described parameters are not intended to be limiting in any way. Furthermore, the mathematical expressions using these parameters may differ from those explicitly disclosed in this disclosure. The various channels (e.g., PUCCH, PDCCH, etc.) and information elements may be identified by any suitable names, and therefore the various names assigned to these various channels and information elements are not intended to be limiting in any way.

[0063] As used in this disclosure, the terms "determining" and "determining" may encompass a wide variety of actions. "Determining" and "determining" may include, for example, judging, calculating, computing, processing, deriving, investigating, looking up, searching, inquiring (e.g., searching in a table, database, or other data structure), ascertaining, and the like. "Determining" and "determining" may also include receiving (e.g., receiving information), transmitting (e.g., sending information), input, output, accessing (e.g., accessing data in memory), and the like. Furthermore, "judgment" and "decision" can include regarding resolving, selecting, choosing, establishing, comparing, etc. as having been "judged" or "decided." In other words, "judgment" and "decision" can include regarding some action as having been "judged" or "decided." Furthermore, "judgment (decision)" can be interpreted as "assuming," "expecting," "considering," etc.

[0064] As used in this disclosure, the phrase "based on" does not mean "based only on," unless expressly stated otherwise. In other words, the phrase "based on" means both "based only on" and "based at least on."

[0065] As used in this disclosure, any reference to an element using a designation such as "first," "second," etc. does not generally limit the quantity or order of those elements. These designations may be used in this disclosure as a convenient method of distinguishing between two or more elements. Thus, a reference to a first and a second element does not imply that only two elements may be employed or that the first element must in some way precede the second element.

[0066] When the terms "include," "including," and variations thereof are used in this disclosure, these terms are intended to be inclusive, similar to the term "comprising." Furthermore, when the term "or" is used in this disclosure, it is not intended to be an exclusive or.

[0067] In this disclosure, where articles are added by translation, such as a, an, and the in English, the disclosure may include that the nouns following these articles are in the plural form.

[0068] In the present disclosure, the term "A and B are different" may mean "A and B are different from each other." The term may also mean "A and B are each different from C." Terms such as "separate" and "coupled" may also be interpreted in the same way as "different."

[0069] 1...sentence generation device, 10...sentence generation model generation device, 11...learning unit, 12...model holding unit, 13...text sentence acquisition unit, M...sentence generation model, 1001...processor, 1002...memory, 1003...storage, 1004...communication device, 1005...input device, 1006...output device, 1007...bus.

Claims

1. A sentence generation model generation device comprising: a learning unit that, in response to input of an input sentence, identifies from the input sentence an assertion and grounds that support the assertion from the same or different standpoint as the assertion, performs at least one of labeling the assertion and the grounds, and labeling the relationship between the assertion and the grounds, and performs machine learning using the input sentence as an explanatory variable and the text of the portion where the labeling was performed as an objective variable, thereby learning a sentence generation model for generating and outputting text sentences related to the structure of the input sentence, thereby generating the sentence generation model.

2. The sentence generation model generating device according to claim 1, wherein the learning unit further identifies, as the claims, a major claim that is a main claim and a normal claim that reinforces the major claim from the same or a different standpoint as the major claim, and performs at least one of labeling the major claim and the normal claim, and labeling the relationship between the major claim and the normal claim.

3. The sentence generation model generating device according to claim 1, wherein the learning unit sets the objective variable in the machine learning to text containing only the identified assertion and evidence.

4. A sentence generation model generation method executed by a sentence generation model generation device, comprising: a step of generating the sentence generation model by conducting learning of a sentence generation model for generating and outputting text sentences related to the structure of an input sentence, the step of generating the sentence generation model including: a sub-step of identifying, from the input sentence, an assertion and grounds that support the assertion from the same position as the assertion or a different position; a sub-step of performing at least one of labeling the assertion and the grounds, and labeling the relationship between the assertion and the grounds; and a sub-step of conducting machine learning using the input sentence as an explanatory variable and the text at the portion where the labeling has been performed as an objective variable.

5. A sentence generation device comprising: a text sentence acquisition unit that acquires a text sentence relating to the structure of a target input sentence as an output from the sentence generation model by inputting a target input sentence into a sentence generation model generated by the sentence generation model generating device described in claim 1.

Images