Device and information processing method

Abstract

To obtain information for more accurately evaluating the importance of a word.SOLUTION: A device 10 comprises: a determination part 11 that determines, for each classification, example sentences where a masked partial portion is replaced with candidate words and the generation probability for each of the candidate words in accordance with the input of text data where the partial portion is masked by using a classification-specific learning model learned using the Masked Language Model learning method based on text data categorized into multiple classifications within a given field; an importance calculation part 12 that calculates the importance of each of the candidate word within each classification based on the example sentences where the masked partial portion is replaced with the candidate words and the generation probability for each of the candidate words determined for each classification; and a basic information acquisition part 13 that acquires basic information to extract important words in each classification based on the calculated importance of each of candidate words within each classification.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present disclosure relates to an apparatus and an information processing method. Note that, in this case, "text" refers to a text in general that includes one or more sentences, and the content of the text is not limited to a specific genre or classification. [Background technology]

[0002] One example is the case where important words (such as unique words or frequently used words) are extracted from a large number of documents related to a certain field, for each predetermined classification or level according to difficulty, etc. (see Patent Document 1). For example, in an English test, there is a need to classify (level) various English words into levels according to difficulty, etc. However, performing such classification work manually takes a great deal of time and is not realistic.

[0003] One method related to automating the classification process is the Term Frequency-Inverse Document Frequency (TF-IDF) method, which calculates a measure of the importance of each word in each document. This method evaluates the importance of a word based on the frequency of its appearance in the document and its rarity in the entire document. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Publication No. 2021-114057 Summary of the Invention [Problem to be solved by the invention]

[0005] However, since the TF-IDF method divides sentences at the word level, it does not take into account the meaning of the words themselves or the context (the relationship between the words before and after them). For example, Example 1: I need to study for my exam tomorrow. Example sentence 2: She conducts a study on effects of climate change. Assuming the above, the word "study" used in Example Sentences 1 and 2 means "the act of studying" in Example Sentence 1, and "research activity" in Example Sentence 2. As such, the word "study" has multiple meanings depending on the context, but the TF-IDF method does not distinguish between these meanings and treats them all as having the same importance. As a result, the TF-IDF method may not be sufficient to evaluate the importance of words with greater accuracy.

[0006] Therefore, an object of the present disclosure is to obtain information for evaluating the importance of words with higher accuracy. [Means for solving the problem]

[0007] The device according to the present disclosure includes: a determination unit that, in response to input of text data with some masked portions, determines, for each category, example sentences in which some masked portions have been replaced with candidate words and the generation probability of each of the candidate words, using a learning model for each category trained using a Masked Language Model learning method based on text data for each of a plurality of categories related to a certain field; an importance calculation unit that calculates the importance of each of the candidate words in each category based on the example sentences in which the masked portions have been replaced with the candidate words and the generation probability of each of the candidate words determined for each category; and a basic information acquisition unit that acquires basic information for extracting important words for each category based on the calculated importance of each of the candidate words in each category. Note that the "Masked Language Model learning method" is one of the pre-training methods for natural language processing models, and is a method of training a model by randomly hiding (masking) words in a document and predicting the identity of the hidden words. [Effects of the Invention]

[0008] According to the present disclosure, information for evaluating the importance of words with higher accuracy can be obtained. [Brief explanation of the drawings]

[0009] [Figure 1] FIG. 2 is a functional block diagram showing the configuration of the device. [Figure 2] FIG. 10 is a flow diagram showing the processing of the learning phase. [Figure 3] FIG. 10 is a flow diagram showing the processing of the inference phase. [Figure 4] FIG. 4 is a diagram for explaining the processing of steps S11 and S12 in FIG. 3. [Figure 5] FIG. 4 is a diagram for explaining the process of step S12 in FIG. 3. [Figure 6] FIG. 4 is a diagram for explaining the process of step S13 in FIG. 3. [Figure 7] FIG. 4 is a diagram for explaining the process of step S14 in FIG. 3. [Figure 8] FIG. 2 illustrates an example of a hardware configuration of the apparatus. DETAILED DESCRIPTION OF THE INVENTION

[0010] Hereinafter, an embodiment of an apparatus and a determination method according to the present disclosure will be described with reference to the drawings.

[0011] Fig. 1 shows a configuration diagram of a device 10 according to the present disclosure. As shown in Fig. 1, in order to realize the functions according to the present disclosure, the device 10 includes a determination unit 11, an importance calculation unit 12, and a basic information acquisition unit 13. The functions of each unit will be outlined below.

[0012] The determination unit 11 is a functional unit that uses a learning model for each category trained using a Masked Language Model learning method based on text data for each of multiple categories related to a certain field to determine, in response to input of text data in which some portions are masked, example sentences in which some masked portions (hereinafter referred to as "masked portions") have been replaced with candidate words and the generation probability of each candidate word for each category. In this embodiment, the field is assumed to be "English exams," and the multiple categories related to the field are assumed to be "multiple levels of English exams (here, advanced, intermediate, and beginner)." Therefore, the determination unit 11 trains learning models M for multiple levels (advanced, intermediate, and beginner) using a Masked Language Model learning method based on text data for each level (advanced, intermediate, and beginner). Furthermore, the determination unit 11 inputs text data in which some portions of the text data for each level are masked into the learning model M for each level, and determines, based on the output from the learning model M, example sentences in which the masked portions have been replaced with candidate words and the generation probability of each candidate word for each level. The learning model M for each level is held by the determination unit 11.

[0013] The importance calculation unit 12 is a functional unit that calculates the importance of each candidate word in each classification based on (1) example sentences in which masked portions have been replaced with candidate words and (2) the probability of generation of each candidate word, which are determined for each classification. Specifically, assuming an English test, the importance calculation unit 12 calculates the importance of each candidate word in each grade based on example sentences in which masked portions for each grade have been replaced with candidate words and the probability of generation of each candidate word.

[0014] The basic information acquisition unit 13 is a functional unit that acquires basic information for extracting important words in each category based on the importance of each candidate word in each category calculated by the importance calculation unit 12. Specifically, assuming an English test, the basic information acquisition unit 13 acquires dictionary data indicating important English words in each level as basic information based on the importance of each candidate word in each level.

[0015] The device 10 may employ various information processing devices (smartphones, mobile phones, smartwatches, wearable devices, notebook computers, desktop computers, servers, etc.) as hardware.

[0016] The processing executed by the device 10 (processing related to the information processing method of the present disclosure) will be described below with reference to the flow diagrams of Figures 2 and 3. The processing executed by the device 10 includes a learning phase processing (Figure 2) for acquiring learning models for each of a plurality of levels (advanced, intermediate, beginner), and an inference phase processing (Figure 3) for acquiring dictionary data for each of a plurality of levels (advanced, intermediate, beginner), which will be described in order below.

[0017] In the learning phase processing shown in Figure 2, the determination unit 11 externally collects a large amount of text data for each of multiple levels (advanced, intermediate, beginner) of an English test (step S1), and uses a Masked Language Model learning method based on the text data for each level to fine-tune a learning model M for each level (e.g., a BERT model) (step S2). As a result, in response to input text data with some masked parts, a learning model M for each level is acquired for outputting example sentences in which the masked parts have been replaced with candidate words and the generation probability for each candidate word. The acquired learning model M for each level is held by the determination unit 11.

[0018] In the inference phase process shown in Fig. 3, the importance calculation unit 12 replaces the word input by the user with [MASK] (step S11). For example, as shown on the left side of Fig. 4, This book is very interesting. is entered, the word "very" will be replaced with [MASK], This book is [MASK] interesting. This becomes:

[0019] Next, the importance calculation unit 12 inputs the replaced sentence into the learning model M for each level and determines, for each level, example sentences in which the masked portions have been replaced with candidate words and the generation probability of each candidate word (step S12). For example, as shown in FIG. 4, the replaced sentence (e.g., "This book is [MASK] interesting.") is input into the learning model M for each level, and multiple example sentences in which the masked portions have been replaced with candidate words are determined based on the output from the learning model M for each level. Then, as shown in FIG. 5, the importance calculation unit 12 calculates generation probabilities using the TF-IDF method from the multiple example sentences output from the learning model M for each level and associates the words with the level. As a result, for the advanced level, the generation probability of the candidate word "very" is determined to be "0.1," the generation probability of the candidate word "extremely" is determined to be "0.3," and the generation probability of the candidate word "phenomenally" is determined to be "0.4." Similarly, the generation probability of each candidate word is determined for the intermediate and beginner levels.

[0020] Next, the importance calculation unit 12 calculates the importance of each candidate word in each class from the generation probability (step S13). For example, as shown in Figure 6, the importance calculation unit 12 calculates the importance of each candidate word in each class using the "generation probability of all candidate words" output from the learning model M of each class for the generation probability of each basic candidate word. At this time, as shown on the right side of Figure 6, in the conventional TF-IDF method, TF(w,d) = number of occurrences of word w in document d / total number of words in document d IDF(w) = log(total number of documents / number of documents containing word w) However, in this embodiment, TF(w,n) = Probability of generating word w for model n IDF(w) = log(sum of all word probabilities for all models / sum of word w probabilities) (Note that the "sum of all word generation probabilities for all models" corresponds to the number of models.) is employed to calculate the importance of each candidate word in each class.

[0021] Furthermore, the basic information acquisition unit 13 acquires dictionary data indicating important English words for each level as basic information based on the importance of each candidate word for each level calculated in step S13 (step S14). For example, in the example of FIG. The importance of the candidate word "very" in the advanced level is 0.146, The importance of the candidate word "very" in the intermediate level is 0.146, The importance of the candidate word "very" in the beginner level is 0.728. Therefore, since the importance of the candidate word "very" in the beginner's level is higher than the importance in the advanced and intermediate levels, the candidate word "very" is determined to be an important English word in the "beginner's level" and is included in the dictionary data for the "beginner's level."

[0022] Similarly, the importance of the candidate word "especially" in the intermediate level is 0.805, The importance of the candidate word "especially" in the beginner level is 0.161, Therefore, since the candidate word "especially" has the highest importance in the intermediate level, the candidate word "especially" is determined to be an important English word in the "intermediate level" and is included in the dictionary data for the "intermediate level." In this way, for each candidate word, it is determined which level of dictionary data it should be included in according to its importance in each level, and dictionary data for each of the advanced, intermediate, and beginner levels is obtained, as shown on the right side of Figure 7. The obtained dictionary data for each level is output to the outside (display output, data output, etc.) in response to, for example, an instruction from the user of device 10.

[0023] According to the embodiment described above, a learning model for each level is trained using a Masked Language Model learning method to output example sentences in which masked portions are replaced with candidate words and the generation probability for each candidate word. Furthermore, using this learning model, the importance of each candidate word for each level is calculated, and dictionary data (basic information for extracting important words) showing important English words for each level is obtained based on the importance. In this way, information (dictionary data for each level) for more accurately evaluating the importance of words can be obtained, taking into account the meaning of the word itself, context (relationship with surrounding words), etc.

[0024] Furthermore, as shown in Figure 6, the importance calculation unit 12 calculates the importance of each candidate word in each class using the "generation probability of all candidate words" output from the learning model M of each class for the generation probability of each candidate word that is the basis for calculating the importance, thereby making it possible to calculate the importance with high accuracy.

[0025] In the above embodiment, an example was described in which the field of application was assumed to be "English exams," but the field of application is not limited to "English exams," and the invention can also be applied to other fields in which multiple classifications (multiple grades) exist.

[0026] The gist of the present disclosure lies in the following [1] to [4]. [1] A determination unit that, in response to input of text data in which some portions are masked, determines, for each category, example sentences in which some masked portions are replaced with candidate words and the generation probability of each of the candidate words, using a learning model for each category that is trained using a Masked Language Model learning method based on text data in each of a plurality of categories related to a certain field; an importance calculation unit that calculates the importance of each of the candidate words in each classification based on the example sentences in which the masked portions are replaced with the candidate words and the generation probabilities of each of the candidate words, determined for each classification; a basic information acquiring unit that acquires basic information for extracting important words in each category based on the calculated importance of each candidate word in each category; An apparatus comprising: [2] The plurality of classifications relating to the field are a plurality of levels relating to an English test; the determination unit determines an example sentence in which the masked portion is replaced with the candidate word and a generation probability of each of the candidate words for each of the plurality of classes; the importance calculation unit calculates the importance of each of the candidate words in each class based on the example sentences and the generation probabilities of each of the candidate words determined for each class; The device described in [1], wherein the basic information acquisition unit acquires dictionary data indicating important English words at each level as the basic information based on the importance of each candidate word at each level. [3] The importance calculation unit calculates the probability of generating each of the candidate words as a basis, The device according to [1] or [2] calculates the importance of each candidate word in each classification using the generation probability of all candidate words output from the learning model for each classification. [4] A step in which the device uses a learning model for each category that has been trained using a Masked Language Model learning method based on text data for each of a plurality of categories related to a certain field, and in response to input of text data in which some parts are masked, determines example sentences in which some masked parts are replaced with candidate words and the generation probability of each of the candidate words for each of the categories; the device calculates the importance of each of the candidate words in each classification based on the example sentences in which the masked portions have been replaced with the candidate words and the generation probabilities of each of the candidate words, determined for each classification; The device acquires basic information for extracting important words in each category based on the calculated importance of each candidate word in each category; An information processing method comprising:

[0027] [Explanation of terms, explanation of hardware configuration (Figure 8), etc.] The block diagrams used to explain 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 for realizing each functional block is not particularly limited. That is, each functional block may be realized using a single device that is physically or logically coupled, or may be realized using two or more physically or logically separated devices that are directly or indirectly connected (for example, using wires, wirelessly, etc.) and these multiple devices. The functional block may also be realized by combining the single device or multiple devices with software.

[0028] Functions include, but are not limited to, judgment, determination, judgment, calculation, computation, processing, derivation, investigation, search, confirmation, reception, transmission, output, access, resolution, selection, election, establishment, comparison, assumption, expectation, consideration, broadcasting, notifying, communicating, forwarding, configuring, reconfiguring, allocation, 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.

[0029] For example, an apparatus according to an embodiment of the present disclosure may function as a computer that executes the processes of the present disclosure. Fig. 8 is a diagram illustrating an example of a hardware configuration of an apparatus 10 according to an embodiment of the present disclosure. The above-described apparatus 10 may be physically configured as a computer 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.

[0030] In the following description, the term "apparatus" can be interpreted as a circuit, a device, a unit, etc. The hardware configuration of apparatus 10 may be configured to include one or more of the apparatuses shown in the drawings, or may be configured to exclude some of the apparatuses.

[0031] Each function of device 10 is realized by loading specified software (programs) onto hardware such as processor 1001 and memory 1002, causing processor 1001 to perform calculations, control communication via communication device 1004, and control at least one of reading and writing data in memory 1002 and storage 1003.

[0032] 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.

[0033] 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-mentioned embodiments. Although 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.

[0034] The memory 1002 is a computer-readable recording medium and may be configured, for example, by 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.

[0035] 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 disc), a smart card, a flash memory (e.g., a card, a stick, a key drive), a floppy 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.

[0036] 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, for example, Frequency Division Duplex (FDD) and Time Division Duplex (TDD).

[0037] The input device 1005 is an input device (for example, 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 (for example, 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 (for example, a touch panel).

[0038] 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.

[0039] Furthermore, device 10 may 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, processor 1001 may be implemented using at least one of these pieces of hardware.

[0040] 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, and 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.

[0041] Each aspect / embodiment described in the present disclosure may be any of the following: 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 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 The present invention may be applied to at least one of systems using 802.20, UWB (Ultra-Wideband), Bluetooth (registered trademark), or other appropriate 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 at least one of LTE and LTE-A with 5G).

[0042] 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.

[0043] 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 sent to another device.

[0044] 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).

[0045] Each aspect / embodiment described in this disclosure may be used alone, in combination, or switched depending on the implementation. Furthermore, notification of predetermined information (e.g., notification that "X is true") is not limited to being done explicitly, but may be done implicitly (e.g., by not notifying the predetermined information).

[0046] 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.

[0047] 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.

[0048] 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.

[0049] 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.

[0050] Note that terms explained 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.

[0051] As used in this disclosure, the terms "system" and "network" are used interchangeably.

[0052] Furthermore, the information, parameters, etc. described in the present disclosure may be expressed using absolute values, may be expressed using relative values ​​from a predetermined value, or may be expressed using other corresponding information. For example, a radio resource may be indicated by an index.

[0053] 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.

[0054] 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.

[0055] 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."

[0056] 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.

[0057] When used in this disclosure, the terms "include," "including," and variations thereof are intended to be inclusive, similar to the term "comprising." Furthermore, when used in this disclosure, the term "or" is not intended to be an exclusive or.

[0058] 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.

[0059] 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." [Explanation of symbols]

[0060] 10...device, 11...determination unit, 12...importance calculation unit, 13...basic information acquisition unit, 1001...processor, 1002...memory, 1003...storage, 1004...communication device, 1005...input device, 1006...output device, 1007...bus.

Claims

1. a determination unit that, in response to input of text data in which some portions are masked, determines, for each category, example sentences in which some masked portions are replaced with candidate words and the generation probability of each of the candidate words, using a learning model for each category that is trained using a Masked Language Model learning method based on text data in each of a plurality of categories related to a certain field; an importance calculation unit that calculates the importance of each of the candidate words in each classification based on the example sentences in which the masked portions are replaced with the candidate words and the generation probabilities of each of the candidate words, determined for each classification; a basic information acquiring unit that acquires basic information for extracting important words in each category based on the calculated importance of each candidate word in each category; An apparatus comprising:

2. the plurality of classifications relating to the field are a plurality of levels relating to an English test; the determination unit determines an example sentence in which the masked portion is replaced with the candidate word and a generation probability of each of the candidate words for each of the plurality of classes; the importance calculation unit calculates the importance of each of the candidate words in each class based on the example sentences and the generation probabilities of each of the candidate words determined for each class; the basic information acquisition unit acquires dictionary data indicating important English words for each level as the basic information based on the importance of each of the candidate words for each level.

10. The apparatus of claim 1.

3. The importance calculation unit calculates the probability of generation of each of the candidate words as a basis by: calculating the importance of each candidate word in each classification using the generation probabilities of all candidate words output from the learning model for each classification; 10. The apparatus of claim 1.

4. a step in which the device uses a learning model for each category that has been trained using a Masked Language Model learning method based on text data for each of a plurality of categories related to a certain field, and in response to input of text data in which some parts are masked, determines example sentences in which some masked parts have been replaced with candidate words and the generation probability of each of the candidate words for each of the categories; the device calculates the importance of each of the candidate words in each classification based on the example sentences in which the masked portions have been replaced with the candidate words and the generation probabilities of each of the candidate words, determined for each classification; The device acquires basic information for extracting important words in each category based on the calculated importance of each candidate word in each category; An information processing method comprising:

Images