Topic recommendation method, device, equipment and storage medium

Abstract

This application provides a method, apparatus, device, and storage medium for recommending topics, including: obtaining words and knowledge points to be tested; inputting the words and knowledge points to be tested into a target model to obtain recommended topics; and recommending the recommended topics to the user. The technical solution provided in this application enables the obtained topics to cover the words and related knowledge points that the user needs to master, helping the user to master the words and related knowledge points efficiently and comprehensively, thereby improving the user experience.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of artificial intelligence (AI) technology, and in particular to a topic recommendation method, apparatus, device, and storage medium. Background Art

[0002] With the continuous advancement of artificial intelligence technology, online learning is becoming more and more popular among students. And consolidating learning outcomes by doing exercises is an important part of the learning process.

[0003] The existing method for setting questions involves the server storing a vast number of questions in a question bank. When a user needs to be tested on a specific word, the server matches the question bank with questions related to that word and recommends them to the user. However, due to the wide variety of question formats and the sheer number of questions, labeling each question with a knowledge point would require significant teaching and research manpower. Consequently, most of the entered questions do not contain any knowledge points. This makes it impossible to assess the knowledge points contained in the user's questions, forcing the user to cover all knowledge points through a massive number of questions. This increases the time and workload for students to practice repetitively, thus reducing the user experience. Summary of the Invention

[0004] The present application provides a topic recommendation method, apparatus, device and storage medium to improve user experience.

[0005] In a first aspect, a question recommendation method is provided, comprising: obtaining words to be tested and knowledge points to be tested; inputting the words to be tested and the knowledge points to be tested into a target model to obtain questions to be recommended; and recommending the questions to be recommended to a user.

[0006] In some possible implementations, before inputting the words and knowledge points to be tested into the target model to obtain the questions to be recommended, the method also includes: obtaining a pre-trained language model and multiple training samples; training the pre-trained language model through the multiple training samples to obtain the target model; wherein each training sample includes: a question, an answer corresponding to the question, and a knowledge point corresponding to the answer.

[0007] In some implementations, the method further includes: determining a knowledge point corresponding to the answer.

[0008] In some implementations, determining the knowledge point corresponding to the answer includes: obtaining a POS tag or a POS tag group corresponding to the answer; and determining the knowledge point corresponding to the answer based on a mapping relationship between the POS tag or the POS tag group corresponding to the answer and the knowledge point.

[0009] In some possible implementations, obtaining a part-of-speech POS tag or a POS tag group corresponding to an answer includes: embedding the answer into a question; inputting the question with the answer into a part-of-speech tagging model to obtain a POS tag for each word in the question; and obtaining a POS tag or a POS tag group corresponding to the answer from the POS tags of each word in the question.

[0010] In some implementations, obtaining a pre-trained language model includes: obtaining a plurality of training topics and an autoregressive language model; and training the autoregressive language model using the plurality of training topics to obtain the pre-trained language model.

[0011] In a second aspect, a question recommendation device is provided, which is characterized in that it includes: a first acquisition module, an input module and a recommendation module, wherein the first acquisition module is used to obtain words to be tested and knowledge points to be tested; the input module is used to input the words to be tested and the knowledge points to be tested into a target model to obtain questions to be recommended; and the recommendation module is used to recommend questions to users.

[0012] In some possible implementations, the device also includes: a second acquisition module and a training module, the second acquisition module is used to acquire a pre-trained language model and multiple training samples; the training module is used to train the pre-trained language model through multiple training samples to obtain a target model; wherein each training sample includes: a question, an answer corresponding to the question, and a knowledge point corresponding to the answer.

[0013] In some implementations, the device further includes: a determination module, configured to determine a knowledge point corresponding to the answer.

[0014] In some implementations, the determination module is specifically used to: obtain a POS tag or a POS tag group corresponding to the answer; and determine the knowledge point corresponding to the answer based on a mapping relationship between the POS tag or the POS tag group corresponding to the answer and the knowledge point.

[0015] In some possible implementations, the determination module is specifically used to: embed the answer into the question; input the question with the answer into a part-of-speech tagging model to obtain POS tags for each word in the question; and obtain a POS tag or POS tag group corresponding to the answer from the POS tags of each word in the question.

[0016] In some implementations, the second acquisition module is specifically used to: acquire multiple training topics and an autoregressive language model; and train the autoregressive language model using the multiple training topics to obtain a pre-trained language model.

[0017] In summary, this application automatically generates questions corresponding to words and knowledge points through a trained target model, so that the questions can cover the words and related knowledge points that users need to master, helping users to master words and related knowledge points efficiently and comprehensively, thereby improving the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0019] Figure 1 A schematic diagram of an application scenario provided in an embodiment of the present application;

[0020] Figure 2 A flowchart of a topic recommendation method provided in an embodiment of the present application;

[0021] Figure 3 A schematic diagram of an interface provided in an embodiment of the present application;

[0022] Figure 4 A schematic diagram of the training process of the target model provided in an embodiment of the present application;

[0023] Figure 5 A schematic diagram of a topic recommendation device 500 provided in an embodiment of the present application;

[0024] Figure 6 It is a schematic block diagram of an electronic device 600 provided in an embodiment of the present application. DETAILED DESCRIPTION

[0025] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0026] It should be noted that the terms "first", "second", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way are interchangeable where appropriate so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0027] As mentioned above, the existing question-generating method involves storing a large number of questions in a question bank. When a user needs to be tested on a specific word, the server matches the question bank with questions related to that word and recommends them to the user. However, due to the wide variety of question formats and the sheer number of questions, labeling each question with a knowledge point consumes a significant amount of teaching and research manpower. Consequently, most of the entered questions do not contain any knowledge points. This makes it impossible to assess the knowledge points contained in the questions the user is asked to complete. The only option is to cover all knowledge points through a massive number of questions, increasing the time and workload for students to practice repetitively, thus reducing the user experience.

[0028] In order to solve this technical problem, this application automatically generates questions corresponding to words and knowledge points through a trained target model, so that the questions can cover the words and related knowledge points that users need to master, helping users to master words and related knowledge points efficiently and comprehensively, thereby improving the user experience.

[0029] The technical solutions of the embodiments of the present application can be applied to the following scenarios, but are not limited thereto: Figure 1 A schematic diagram of an application scenario provided in an embodiment of the present application is shown in FIG. Figure 1 As shown, the terminal device 110 can communicate with the server 120. The terminal device 110 can be installed with an online learning application (APP), and the server 120 can implement the background services corresponding to the APP. For example, a user can enter a word or knowledge point to be tested in the APP. The server 120 can process the word or knowledge point in the background, generate questions related to the word or knowledge point, and recommend them to the user.

[0030] In some implementations, the terminal device may be a smartphone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a wearable device, a learning machine, etc., but is not limited thereto.

[0031] In some implementations, the server may be one or more physical servers, or a cloud server providing cloud computing services, which is not limited in this application.

[0032] It should be understood that Figure 1 It only includes one terminal device and one server, which is just an example. In fact, the application scenario may include multiple terminal devices, multiple servers, etc. For example, one server can communicate with multiple terminal devices.

[0033] The technical solutions of the embodiments of the present application are described in detail below:

[0034] Figure 2 This is a flowchart of a topic recommendation method provided in an embodiment of the present application. The method can be applied to Figure 1 On the server 120 shown, but not limited to this, as Figure 2 As shown, the method includes the following process:

[0035] S210: Obtain the words and knowledge points to be tested.

[0036] S220: Input the words to be tested and the knowledge points to be tested into the target model to obtain the recommended questions.

[0037] S230: Recommending topics to be recommended to the user.

[0038] In some implementations, the user can use the terminal device, such as Figure 1 Open the online learning APP on the terminal device 1 shown in the figure, as shown in the figure Figure 3 As shown, enter the words to be tested and the knowledge points to be tested in the input box of the APP and press the Enter key so that the server can obtain the words to be tested and the knowledge points to be tested. Alternatively, the user can input the words to be tested and the knowledge points to be tested by voice or other means, so that the online learning APP can obtain the words to be tested and the knowledge points to be tested, and send the words to be tested and the knowledge points to be tested to the server. In short, this application does not limit the way the server obtains the words to be tested and the knowledge points to be tested.

[0039] It should be understood that the input of the above target model is words and knowledge points, and the output is questions related to the words and knowledge points. When the server obtains the words and knowledge points to be tested, the words and knowledge points to be tested can be input into the target model to obtain recommended questions, that is, questions related to the words and knowledge points to be tested.

[0040] Example 1: Assume that the word entered by the user is "go" and the knowledge point to be tested is "past tense". The server can recommend a fill-in-the-blank question: "HE_to the zoo yesterday", or a multiple-choice question: "HE_to thezoo yesterday" "A: went B: is going to C: go D: going to".

[0041] Example 2: Assume that the word entered by the user is "go" and the knowledge point to be tested is "be + present participle and preposition". The server can recommend a fill-in-the-blank question to the user: "HE_to the zoo next week" or a multiple-choice question: "HE_to the zoo next week" with "A: went B: is going to C: go D: going to".

[0042] This application can automatically generate questions corresponding to the words and knowledge points to be tested through the target model, so that the questions can cover the words and related knowledge points that users need to master, helping users to master the words and related knowledge points efficiently and comprehensively, thereby improving the user experience.

[0043] The following describes the process of obtaining the target model:

[0044] In some implementations, the server can use multiple training questions to train an autoregressive language model to obtain a pre-trained language model. Then, the server can train the pre-trained language model with multiple training samples to obtain a target model, such as Figure 4 Alternatively, the server can create an initial model whose inputs are words and knowledge points, and whose outputs are questions related to the words and knowledge points. In other words, the inputs and outputs of the initial model are identical to those of the target model, except that the parameters of the initial model are different from those of the target model. Furthermore, the server can train the initial model using multiple training samples to obtain the target model.

[0045] It should be understood that the above-mentioned multiple training questions are all untagged training questions, that is, training questions without part-of-speech tags. For example, primary and secondary school English questions can be used as training questions.

[0046] It should be understood that predicting the next possible word based on the previous content is what is often called a left-to-right language model task, or vice versa, that is, predicting the previous word based on the following content. This type of model is called an autoregressive language model.

[0047] In some implementations, the autoregressive language model may be a model combining Bidirectional and Auto-Regressive Transformers (BART) model, a text-to-text transfer transformer (T5) model, etc., but is not limited thereto.

[0048] BART is a denoising autoencoder that establishes a mapping from a damaged document to the original document. It uses a sequence-to-sequence (seq2seq) model that includes a bidirectional encoder and a left-to-right autoregressive decoder. During pre-training, BART destroys the text and then optimizes the reconstruction loss.

[0049] The basic idea of the T5 model is to view each Natural Language Processing (NLP) problem as a text-to-text problem, that is, taking text as input and generating new text as output. This allows the same model, objectives, training steps, and decoding process to be directly applied to each task.

[0050] It should be understood that the above-mentioned pre-trained language model is also an autoregressive language model, which is the basis for subsequent question generation and can effectively establish information such as word meaning, grammar, and question structure.

[0051] It should be understood that each of the above training samples includes: a question, an answer corresponding to the question, and a knowledge point corresponding to the answer.

[0052] In some implementable methods, when training a pre-trained language model, the answer corresponding to the question and the knowledge point corresponding to the answer can be input into the pre-trained language model to obtain a predicted question, and the loss of the pre-trained language model is calculated based on the predicted question and the actual question in the training sample. When the loss is less than a preset value, the training of the pre-trained language model is stopped. Alternatively, the server determines whether the number of training times for the pre-trained language model has reached a preset number, and if so, the training of the pre-trained language model is stopped. Alternatively, after training the pre-trained language model through the above-mentioned multiple training samples, the server obtains the parameters of the trained pre-trained language model, and the server determines whether the parameters converge, and if so, stops training the pre-trained language model. In short, this application does not impose any restrictions on the training method of the pre-trained language model, and does not impose any restrictions on the training stop conditions.

[0053] It should be understood that the input of the above target model is words and knowledge points, and the output is questions related to the words and knowledge points.

[0054] It should be understood that the knowledge point corresponding to the answer in each of the above training samples can be obtained by any of the following feasible methods, but is not limited thereto:

[0055] In one implementable manner, the server may obtain a Parts Of Speech (POS) tag or a POS tag group corresponding to the answer, and determine the knowledge point corresponding to the answer based on a mapping relationship between the POS tag or POS tag group corresponding to the answer and the knowledge point.

[0056] In another possible implementation, the server can directly input the answer into a neural network model, which can then obtain the knowledge point corresponding to the answer. The neural network model can be obtained by using a large number of training samples, each of which includes: the answer and the knowledge point corresponding to the answer.

[0057] The following is a detailed explanation of the first implementation method:

[0058] It should be understood that each word corresponds to a POS tag, and if an answer includes multiple words, the POS tags corresponding to all words in the answer constitute a POS tag group.

[0059] Example 3: Suppose the answer to a question is went, which is a word, and its corresponding POS tag is a POS tag, such as "VBD", which represents the past tense of a verb.

[0060] Example 4: Suppose the answer to a question is "is going to." This is a three-word answer, so the corresponding POS tag group is: "VBZ" for "is," indicating the third-person singular verb; "VBG" for "going," indicating the verb participle; and "IN" for "to," indicating the preposition. Therefore, the corresponding POS tag group for this answer is: "VBZ," "VBG," and "IN."

[0061] The server can embed the answer into the question, input the question with the answer into the part-of-speech tagging model to obtain the POS tags of each word in the question, and obtain the POS tag or POS tag group corresponding to the answer from the POS tags of each word in the question.

[0062] It should be understood that the above-mentioned part-of-speech tagging model can be any existing part-of-speech tagging model, and this application does not impose any limitation on this.

[0063] In some implementations, the server may establish mapping relationships between some high-frequency POS tags or POS tag groups and knowledge points.

[0064] For example, Table 1 shows the mapping relationship between some high-frequency POS tags or POS tag groups and knowledge points, as shown in Table 1:

[0065] Table 1

[0066]

[0067]

[0068]

[0069] Furthermore, after obtaining the mapping relationship between the high-frequency POS tags or POS tag groups and the knowledge points, the server can search for the knowledge points corresponding to the POS tags or POS tag groups corresponding to the answers by looking up a table.

[0070] Example 5: Suppose the answer to a question is went, which is a word, and its corresponding POS tag is "VBD". By looking up Table 1, we can know that the knowledge point corresponding to this tag is: past tense of verbs.

[0071] Example 6: Suppose the answer to a question is "is going to." This is a three-word answer, so the corresponding POS tag group is: "VBZ" for "is," indicating the third-person singular verb; "VBG" for "going," indicating the verb participle; and "IN" for "to," indicating the infinitive. The corresponding POS tag group for this answer is: "VBZ," "VBG," and "IN." Table 1 shows that the knowledge points corresponding to these tags are: "be" + present participle and preposition.

[0072] In this application, the server can establish a mapping relationship between some high-frequency POS tags or POS tag groups and knowledge points, based on which the server can search for the knowledge points corresponding to the answers.

[0073] Figure 5 A schematic diagram of a topic recommendation device 500 provided in an embodiment of the present application, wherein the topic recommendation device 500 may be as follows Figure 1 The server 120 shown, but not limited to, Figure 5As shown, the topic recommendation device 500 includes: a first acquisition module 510, an input module 520 and a recommendation module 530, wherein the first acquisition module 510 is used to obtain the words to be tested and the knowledge points to be tested; the input module 520 is used to input the words to be tested and the knowledge points to be tested into the target model to obtain topics to be recommended; the recommendation module 530 is used to recommend topics to users.

[0074] In some possible implementations, the question recommendation device 500 also includes: a second acquisition module 540 and a training module 550, the second acquisition module 540 is used to obtain a pre-trained language model and multiple training samples; the training module 550 is used to train the pre-trained language model through multiple training samples to obtain a target model; wherein each training sample includes: a question, an answer corresponding to the question, and a knowledge point corresponding to the answer.

[0075] In some implementations, the question recommendation device 500 further includes: a determination module 560, configured to determine the knowledge point corresponding to the answer.

[0076] In some implementations, the determination module 560 is specifically configured to: obtain a POS tag or a POS tag group corresponding to the answer; and determine the knowledge point corresponding to the answer based on a mapping relationship between the POS tag or the POS tag group corresponding to the answer and the knowledge point.

[0077] In some implementations, the determination module 560 is specifically used to: embed the answer into the question; input the question with the answer into a part-of-speech tagging model to obtain POS tags for each word in the question; and obtain the POS tag or POS tag group corresponding to the answer from the POS tags of each word in the question.

[0078] In some implementations, the second acquisition module 540 is specifically used to: acquire a plurality of training topics and an autoregressive language model; and train the autoregressive language model using the plurality of training topics to obtain a pre-trained language model.

[0079] It should be understood that the device embodiment and the method embodiment may correspond to each other, and similar descriptions may refer to the method embodiment. To avoid repetition, they will not be described here. Specifically, Figure 5 The apparatus 500 shown may perform Figure 2 The corresponding method embodiments, and the aforementioned and other operations and / or functions of each module in the apparatus 500 are respectively to implement Figure 2 For the sake of brevity, the corresponding processes in each method are not repeated here.

[0080] The above describes the device 500 of the embodiment of the present application from the perspective of functional modules in conjunction with the accompanying drawings. It should be understood that the functional module can be implemented in hardware form, can be implemented by instructions in software form, and can also be implemented by a combination of hardware and software modules. Specifically, the steps of the method embodiment in the embodiment of the present application can be completed by the hardware integrated logic circuit and / or software form instructions in the processor, and the steps of the method disclosed in the embodiment of the present application can be directly embodied as being executed by a hardware decoding processor, or can be executed by a combination of hardware and software modules in the decoding processor. Optionally, the software module can be located in a mature storage medium in the art such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, etc. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps in the above method embodiment in conjunction with its hardware.

[0081] Figure 6 600 is a schematic block diagram of an electronic device 600 provided in an embodiment of the present application. The electronic device 600 may be as follows Figure 1 Server 120 is shown, but not limited to.

[0082] like Figure 6 As shown, the electronic device 600 may include:

[0083] The memory 610 and the processor 620 are configured to store computer programs and transmit the program code to the processor 620. In other words, the processor 620 can call and run the computer program from the memory 610 to implement the method in the embodiment of the present application.

[0084] For example, the processor 620 may be configured to execute the above method embodiments according to instructions in the computer program.

[0085] In some embodiments of the present application, the processor 620 may include but is not limited to:

[0086] General-purpose processor, digital signal processor (DSP), application-specific integrated circuit (ASIC), field-programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware components, etc.

[0087] In some embodiments of the present application, the memory 610 includes but is not limited to:

[0088] Volatile memory and / or non-volatile memory. Non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory can be random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link DRAM (SLDRAM), and direct RAM bus random access memory (DR RAM).

[0089] In some embodiments of the present application, the computer program may be divided into one or more modules, which are stored in the memory 610 and executed by the processor 620 to implement the method provided by the present application. The one or more modules may be a series of computer program instruction segments capable of implementing specific functions, and the instruction segments are used to describe the execution process of the computer program in the electronic device.

[0090] like Figure 6 As shown, the electronic device may further include:

[0091] The transceiver 630 may be connected to the processor 620 or the memory 610 .

[0092] The processor 620 may control the transceiver 630 to communicate with other devices. Specifically, the processor 620 may send information or data to other devices or receive information or data sent by other devices. The transceiver 630 may include a transmitter and a receiver. The transceiver 630 may further include one or more antennas.

[0093] It should be understood that the various components in the electronic device are connected via a bus system, wherein the bus system includes not only a data bus but also a power bus, a control bus and a status signal bus.

[0094] The present application also provides a computer storage medium having a computer program stored thereon, which, when executed by a computer, enables the computer to perform the method of the above-mentioned method embodiment. In other words, the present application also provides a computer program product containing instructions, which, when executed by a computer, enables the computer to perform the method of the above-mentioned method embodiment.

[0095] When software is used for implementation, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function according to the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center via a wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) method. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a digital video disc (DVD)), or a semiconductor medium (e.g., a solid state drive (SSD)).

[0096] Those skilled in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0097] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the modules is merely a logical function division. In actual implementation, there may be other division methods, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0098] Modules described as separate components may or may not be physically separate, and components displayed as modules may or may not be physical modules, i.e., they may be located in one place or distributed across multiple network elements. Some or all of the modules may be selected based on actual needs to achieve the purpose of the present embodiment. For example, the functional modules in the various embodiments of the present application may be integrated into a processing module, or each module may exist physically separately, or two or more modules may be integrated into a single module.

[0099] The above is only a specific embodiment of the present application, but the scope of protection of the present application is not limited to this. Any changes or substitutions that can be easily conceived by any person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A topic recommendation method, characterized in that: include: Get the words and knowledge points to be tested; Obtaining a pre-trained language model and a plurality of training samples, wherein the pre-trained language model is obtained by training an autoregressive language model with a plurality of training questions; Each of the training samples includes: a question, an answer corresponding to the question, and a knowledge point corresponding to the answer; The pre-trained language model is trained using the plurality of training samples to obtain a target model; wherein the training process stops when any of the following conditions is met: the loss of the pre-trained language model is less than a preset value; the number of training times for the pre-trained language model reaches a preset number; or the parameters of the pre-trained language model converge; Inputting the words to be tested and the knowledge points to be tested into the target model to obtain recommended questions corresponding to the words to be tested and the knowledge points to be tested, so that the recommended questions cover the words and related knowledge points that the user needs to master, and the recommended questions include fill-in-the-blank questions and / or multiple-choice questions; if the same word to be tested and different knowledge points to be tested are input into the target model, different recommended questions corresponding to the same word to be tested and different knowledge points to be tested are obtained; Recommending the topic to be recommended to the user; Determining the knowledge point corresponding to the answer includes: obtaining a part-of-speech POS tag or POS tag group corresponding to the answer; obtaining a pre-constructed mapping relationship table of high-frequency POS tags or POS tag groups and knowledge points; and determining the knowledge point corresponding to the answer based on the mapping relationship between the POS tag or POS tag group corresponding to the answer and the knowledge point by searching the mapping relationship table.

2. The method according to claim 1, characterized in that The obtaining of a part-of-speech POS tag or a POS tag group corresponding to the answer includes: embedding the answer into the question; Inputting the question with the answer into a part-of-speech tagging model to obtain POS tags for each word in the question; The POS tag or POS tag group corresponding to the answer is obtained from the POS tags of each word in the question.

3. The method according to claim 1 or 2, characterized in that Obtaining the pre-trained language model includes: Get multiple training questions and autoregressive language models; The autoregressive language model is trained using the multiple training questions to obtain the pre-trained language model.

4. A topic recommendation device, characterized in that: include: The first acquisition module is used to acquire the words and knowledge points to be tested; A second acquisition module is used to acquire a pre-trained language model and a plurality of training samples, wherein the pre-trained language model is obtained by training an autoregressive language model with a plurality of training questions; Each of the training samples includes: a question, an answer corresponding to the question, and a knowledge point corresponding to the answer; A training module, configured to train the pre-trained language model using the plurality of training samples to obtain a target model; wherein the training process stops when any of the following conditions is met: the loss of the pre-trained language model is less than a preset value; the number of training times for the pre-trained language model reaches a preset number; or the parameters of the pre-trained language model converge; An input module is used to input the words to be tested and the knowledge points to be tested into a target model to obtain recommended questions corresponding to the words to be tested and the knowledge points to be tested, so that the recommended questions cover the words and related knowledge points that the user needs to master, and the recommended questions include fill-in-the-blank questions and / or multiple-choice questions; if the same word to be tested and different knowledge points to be tested are input into the target model, different recommended questions corresponding to the same word to be tested and different knowledge points to be tested are obtained; A recommendation module, configured to recommend the topic to be recommended to the user; A determination module is used to determine the knowledge point corresponding to the answer, including: obtaining a part-of-speech POS tag or POS tag group corresponding to the answer; obtaining a pre-built mapping relationship table of high-frequency POS tags or POS tag groups and knowledge points; and determining the knowledge point corresponding to the answer based on the mapping relationship between the POS tag or POS tag group corresponding to the answer and the knowledge point by searching the mapping relationship table.

5. The device according to claim 4, characterized in that The determining module is specifically configured to: embedding the answer into the question; Inputting the question with the answer into a part-of-speech tagging model to obtain POS tags for each word in the question; The POS tag or POS tag group corresponding to the answer is obtained from the POS tags of each word in the question.

6. The device according to claim 4 or 5, characterized in that The second acquisition module is specifically used for: Get multiple training questions and autoregressive language models; The autoregressive language model is trained using the multiple training questions to obtain the pre-trained language model.

7. An electronic device, characterized in that: include: A processor and a memory, the memory being used to store a computer program, and the processor being used to call and run the computer program stored in the memory to execute the method according to any one of claims 1 to 3.

8. A computer-readable storage medium, characterized in that Used to store a computer program, wherein the computer program causes a computer to execute the method according to any one of claims 1 to 3.

Images