News identification method and device, equipment and storage medium

Abstract

The invention relates to the field of artificial intelligence and financial science and technology, and discloses a news recognition method, device and equipment and a storage medium, and the method comprises the steps: obtaining a preset recognition result corresponding to preset news, and obtaining financial professional knowledge corresponding to a financial platform; performing feature extraction on the preset news to obtain a first feature vector, and performing feature extraction on the financial professional knowledge to obtain a second feature vector; fusing the first feature vector and the second feature vector to generate a third feature vector; obtaining a prediction recognition result output by the pre-training language model based on the third feature vector, and generating a loss value between the prediction recognition result and a preset recognition result; when the loss value is smaller than a preset value, obtaining a pre-training language model passing the test; and inputting the current news into the pre-training language model passing the test, and obtaining a current recognition result output by the pre-training language model passing the test based on the current news. The method is beneficial for improving the recognition efficiency of the current news.

Description

Technical Field

[0001] The present invention relates to the fields of artificial intelligence and financial technology, and in particular to a news recognition method, device, equipment and storage medium. Background Art

[0002] The news sources on financial platforms include official publishing agencies, financial media, financial research institutions and self-media institutions. These news cover the dynamics of various financial markets as well as information on macroeconomic policies, international economic relations, etc.

[0003] However, the identification process of the current news of the financial platform is cumbersome, which is not conducive to improving the recognition efficiency of the current news. The reason is that the existing technology mainly adopts the manual recognition method to identify the current news of the financial platform, and the manual recognition method consumes a lot of human resources and time resources, increases the recognition time of the current news of the financial platform, and is easily affected by manual intervention. Therefore, it is not conducive to improving the recognition efficiency of the current news. Summary of the invention

[0004] The present invention provides a news recognition method, device, computer equipment and storage medium to solve the technical problem that the recognition process of current news on the existing financial platform is complicated and is not conducive to improving the recognition efficiency of current news.

[0005] In a first aspect, a news identification method is provided, comprising:

[0006] Acquire confirmed false news and real news on the financial platform, select the false news or the real news as preset news, obtain preset recognition results corresponding to the preset news, and obtain financial expertise corresponding to the financial platform;

[0007] Extracting features from the preset news to obtain a first feature vector, and extracting features from the financial expertise to obtain a second feature vector;

[0008] Merging the first eigenvector and the second eigenvector to generate a third eigenvector;

[0009] Obtaining a predicted recognition result output by a pre-trained language model based on the third feature vector, and generating a loss value between the predicted recognition result and the preset recognition result based on a predefined generation method;

[0010] When the loss value is less than a preset value, a predefined acquisition method is used to acquire the pre-trained language model that has passed the test;

[0011] The current news of the financial platform is obtained, the current news is input into the pre-trained language model that has passed the test, and the current recognition result output by the pre-trained language model that has passed the test based on the current news is obtained.

[0012] Furthermore, the feature extraction of the preset news to obtain a first feature vector and the feature extraction of the financial expertise to obtain a second feature vector include:

[0013] Acquire news text, news image, and news audio of the preset news, perform feature extraction on the news text, news image, and news audio of the preset news, and obtain a first feature vector;

[0014] Financial terms and financial regulations in the financial expertise are obtained, and features of the financial terms and financial regulations are extracted to obtain a second feature vector.

[0015] Further, the fusing the first feature vector and the second feature vector to generate a third feature vector includes:

[0016] Using a linear transformation layer of the pre-trained language model, convert the first feature vector into a first output vector, and convert the second feature vector into a second output vector;

[0017] The first output vector and the second output vector are fused through a multi-head self-attention mechanism to generate a third feature vector.

[0018] Further, the obtaining of the predicted recognition result output by the pre-trained language model based on the third feature vector and generating a loss value between the predicted recognition result and the preset recognition result based on a predefined generation method include:

[0019] Obtaining a predicted recognition result output by a classifier of a pre-trained language model based on the third feature vector;

[0020] A loss value between the predicted recognition result and the preset recognition result is generated through a preset cross entropy loss function.

[0021] Furthermore, when the loss value is less than a preset value, a predefined acquisition method is adopted to acquire the pre-trained language model that has passed the test, including:

[0022] When the loss value is less than a preset value, obtaining a test set corresponding to the financial platform, using the test set to test the pre-trained language model, and generating a test result;

[0023] According to the test result, the current accuracy of the pre-trained language model on the test set is obtained, and when the current accuracy is greater than a preset accuracy, the pre-trained language model that passes the test is obtained.

[0024] Furthermore, the obtaining of the current news of the financial platform, inputting the current news into the pre-trained language model that has passed the test, and obtaining the current recognition result output by the pre-trained language model that has passed the test based on the current news, includes:

[0025] Acquire a chat interface of the financial platform, acquire a dialogue window of the chat interface, and acquire the current news appearing in the dialogue window;

[0026] Acquire news text, news image, and news audio of the current news, perform feature extraction on the news text, news image, and news audio of the current news to obtain a fourth feature vector, input the fourth feature vector into the pre-trained language model that has passed the test, and obtain a current recognition result output by the pre-trained language model that has passed the test based on the fourth feature vector.

[0027] Furthermore, after obtaining the current news of the financial platform, inputting the current news into the pre-trained language model that has passed the test, and obtaining the current recognition result output by the pre-trained language model that has passed the test based on the current news, the news recognition method includes:

[0028] A prompt box of the financial platform is obtained, and the current recognition result is displayed through the prompt box.

[0029] In a second aspect, a news identification device is provided, comprising:

[0030] The first acquisition module is used to acquire confirmed false news and real news on the financial platform, select the false news or the real news as preset news, acquire preset recognition results corresponding to the preset news, and acquire financial expertise corresponding to the financial platform;

[0031] An extraction module, used for performing feature extraction on the preset news to obtain a first feature vector, and performing feature extraction on the financial expertise to obtain a second feature vector;

[0032] A generating module, used for fusing the first feature vector and the second feature vector to generate a third feature vector;

[0033] A second acquisition module is used to obtain a predicted recognition result output by a pre-trained language model based on the third feature vector, and generate a loss value between the predicted recognition result and the preset recognition result based on a predefined generation method;

[0034] A third acquisition module is used to acquire the pre-trained language model that passes the test by adopting a predefined acquisition method when the loss value is less than a preset value;

[0035] The recognition module is used to obtain the current news of the financial platform, input the current news into the pre-trained language model that has passed the test, and obtain the current recognition result output by the pre-trained language model that has passed the test based on the current news.

[0036] In a third aspect, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the steps of the above-mentioned news identification method are implemented when the processor executes the computer program.

[0037] In a fourth aspect, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned news identification method are implemented.

[0038] The present application provides a news identification method, apparatus, computer equipment and storage medium, which obtain confirmed false news and real news on a financial platform, select the false news or the real news as preset news, obtain a preset identification result corresponding to the preset news, and obtain the financial expertise corresponding to the financial platform; perform feature extraction on the preset news to obtain a first feature vector, perform feature extraction on the financial expertise to obtain a second feature vector; fuse the first feature vector and the second feature vector to generate a third feature vector; obtain a predicted identification result output by a pre-trained language model based on the third feature vector, and generate a loss value between the predicted identification result and the preset identification result based on a predefined generation method; when the loss value is less than a preset value, use a predefined acquisition method to obtain the predicted identification result. Formula, obtain the pre-trained language model that has passed the test; obtain the current news of the financial platform, input the current news into the pre-trained language model that has passed the test, and obtain the current recognition result output by the pre-trained language model that has passed the test based on the current news. The beneficial effects are in two aspects. On the one hand, the current news of the financial platform is recognized, the current news is input into the pre-trained language model that has passed the test, and the current recognition result output by the pre-trained language model that has passed the test based on the current news is recognized. Since no manual recognition is required, the recognition time of the current news of the financial platform is reduced, which is beneficial to improving the recognition efficiency of the current news of the financial platform; on the other hand, since the pre-trained language model will not be affected by manual intervention, it is beneficial to improve the reliability of the current recognition result. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0040] Figure 1 is a schematic diagram of an application environment of a news identification method in an embodiment of the present invention;

[0041] Figure 2 A flowchart of a news identification method provided by an embodiment of the present invention;

[0042] Figure 3 yes Figure 1 A schematic flow chart of a specific implementation of step S23;

[0043] Figure 4 yes Figure 1 A schematic flow chart of a specific implementation of step S25;

[0044] Figure 5 yes Figure 1 A schematic flow chart of a specific implementation of step S26;

[0045] Figure 6 is a structural schematic diagram of a news identification device in one embodiment of the present invention;

[0046] Figure 7 is a schematic diagram of a structure of a computer device in one embodiment of the present invention;

[0047] Figure 8 It is another structural schematic diagram of a computer device in one embodiment of the present invention. DETAILED DESCRIPTION

[0048] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are 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 creative work are within the scope of protection of the present invention.

[0049] See also Figure 1 , Figure 1 FIG. 1 is a schematic diagram of an application environment of a news identification method according to an embodiment of the present invention. The news identification method provided by the embodiment of the present invention can be applied in the following embodiments: Figure 1 In an application environment, a client communicates with a server through a network.

[0050] The server obtains confirmed false news and real news on the financial platform through the client, selects the false news or the real news as preset news, obtains preset recognition results corresponding to the preset news, and obtains financial expertise corresponding to the financial platform;

[0051] Extracting features from the preset news to obtain a first feature vector, and extracting features from the financial expertise to obtain a second feature vector;

[0052] Merging the first eigenvector and the second eigenvector to generate a third eigenvector;

[0053] Obtaining a predicted recognition result output by a pre-trained language model based on the third feature vector, and generating a loss value between the predicted recognition result and the preset recognition result based on a predefined generation method;

[0054] When the loss value is less than a preset value, a predefined acquisition method is used to acquire the pre-trained language model that has passed the test;

[0055] The current news of the financial platform is obtained, the current news is input into the pre-trained language model that has passed the test, and the current recognition result output by the pre-trained language model that has passed the test based on the current news is obtained.

[0056] In the scheme implemented by the above-mentioned news recognition method, device, equipment and medium, the beneficial effects are in two aspects. On the one hand, the current news of the financial platform is identified, and the current news is input into the pre-trained language model that has passed the test. The pre-trained language model that has passed the recognition test outputs a current recognition result based on the current news. Since no manual recognition is required, the recognition time of the current news of the financial platform is reduced, which is beneficial to improving the recognition efficiency of the current news of the financial platform. On the other hand, since the pre-trained language model will not be affected by manual intervention, it is beneficial to improve the reliability of the current recognition result.

[0057] The client may include but is not limited to various personal computers, laptops, smart phones, tablet computers, and portable wearable devices.

[0058] The server can be implemented with an independent task database or a task database cluster composed of multiple task databases. The present invention is described in detail below through specific embodiments.

[0059] See also Figure 2 , Figure 2 A flowchart of a news identification method provided by an embodiment of the present invention includes the following steps:

[0060] S21, obtaining confirmed false news and true news on the financial platform, selecting the false news or the true news as preset news, obtaining a preset recognition result corresponding to the preset news, and obtaining financial expertise corresponding to the financial platform;

[0061] Among them, financial platforms refer to networked and digital platforms that provide financial services or promote financial transactions. Financial platforms use modern information technologies such as the Internet, mobile Internet, big data, and cloud computing to connect the supply and demand sides of funds and achieve effective allocation and efficient circulation of financial resources.

[0062] Among them, financial platforms include banking platforms, securities platforms, and insurance platforms.

[0063] Exemplarily, the obtaining of confirmed false news and true news on the financial platform, selecting the false news or the true news as preset news, obtaining a preset recognition result corresponding to the preset news, and obtaining financial expertise corresponding to the financial platform include:

[0064] Get verified fake and real news on financial platforms;

[0065] The real news and the real label form a real sample, and the false news and the false label form a false sample;

[0066] Combining different real samples and different false samples into a training set;

[0067] When any one of the false news in the training set is selected as the preset news, the false label is selected as the preset recognition result;

[0068] When any one of the real news in the training set is selected as the preset news, the real label is selected as the preset recognition result;

[0069] Obtain financial expertise corresponding to the financial platform.

[0070] S22, extracting features from the preset news to obtain a first feature vector, and extracting features from the financial expertise to obtain a second feature vector;

[0071] The step of extracting features from the preset news to obtain a first feature vector and extracting features from the financial expertise to obtain a second feature vector includes:

[0072] Acquire news text, news image, and news audio of the preset news, perform feature extraction on the news text, news image, and news audio of the preset news, and obtain a first feature vector;

[0073] Financial terms and financial regulations in the financial expertise are obtained, and features of the financial terms and financial regulations are extracted to obtain a second feature vector.

[0074] Financial terminology is the professional vocabulary and phrases used in the financial field to describe and explain various financial activities, tools, markets, strategies, and risks. Financial terminology covers all aspects of finance, including investment, financing, risk management, financial markets, financial institutions, and financial instruments.

[0075] Among them, financial regulations are the general term for laws, regulations and policies formulated by the state to regulate the behavior of financial markets and financial institutions. Financial regulations are an important tool for maintaining financial market order, protecting consumer rights and interests, and preventing financial risks.

[0076] Among them, news text, news images, and news audio can be distinguished and classified according to the file format suffix. For example, news audio is in mp3 format, news images are in jpg and png formats, and news text documents are in PDF, DOCX, and DOC formats.

[0077] Among them, the Chinese name of MP3 format: MP3 audio format, the English name of MP3 format: MPEG Audio LayerIII, MP3 is an audio compression technology, its full name is Moving Picture Experts Compression Standard Audio Layer 3, abbreviated as MP3.

[0078] Among them, the Chinese name of JPG: Joint Photographic Experts Group format, the English name of JPG: Joint Photographic Experts Group, JPG is a widely used image compression format, especially suitable for the compression of photos and complex images.

[0079] Among them, the Chinese name of PNG: Portable Network Graphics Format, and the English name of PNG: Portable Network Graphics.

[0080] Among them, the Chinese name of PDF: Portable Document Format, the English name of PDF: Portable Document Format. PDF files are highly portable and reliable, and can be viewed and printed on any device without losing the original format and layout.

[0081] Among them, the English name of DOCX: Open XML Document, DOCX is an extensible way to store text and format information, making the organization and management of documents more flexible and efficient.

[0082] Among them, DOC is the abbreviation of Document, which is a file extension used to represent a specific document format.

[0083] Exemplarily, the news text of the preset news includes the title, author, release date, news body, and keywords of the preset news.

[0084] S23, fusing the first feature vector and the second feature vector to generate a third feature vector;

[0085] S24, obtaining a predicted recognition result output by the pre-trained language model based on the third feature vector, and generating a loss value between the predicted recognition result and the preset recognition result based on a predefined generation method;

[0086] Among them, the pre-trained language model (PLM) is a language model pre-trained on large-scale text data.

[0087] Among them, the pre-trained language model is trained on large-scale text data, so it can learn rich language knowledge and representation. This knowledge transfer ability enables the model to converge faster when adapting to new tasks and achieve better performance in natural language processing tasks.

[0088] Preferably, the pre-trained language model is a Transformer model, the Transformer model has 24 layers, 1024 hidden layer units, 16 attention heads, and each attention head has 64 hidden layer units.

[0089] Among them, the Chinese name of the Transformer model is converter model.

[0090] The obtaining of the predicted recognition result output by the pre-trained language model based on the third feature vector and generating a loss value between the predicted recognition result and the preset recognition result based on a predefined generation method include:

[0091] Obtaining a predicted recognition result output by a classifier of a pre-trained language model based on the third feature vector;

[0092] A loss value between the predicted recognition result and the preset recognition result is generated through a preset cross entropy loss function.

[0093] S25, when the loss value is less than a preset value, using a predefined acquisition method to acquire the pre-trained language model that has passed the test;

[0094] Exemplarily, when the loss value is less than a preset value, a predefined acquisition method is used to acquire the pre-trained language model that has passed the test, including:

[0095] Training the pre-trained language model based on the loss value, and when the loss value is less than a preset value, stopping the training of the pre-trained language model, and acquiring the pre-trained language model after training;

[0096] Obtaining a test set corresponding to the financial platform, using the test set to test the pre-trained language model after training, and generating a test result;

[0097] According to the test result, the current accuracy of the pre-trained language model after training on the test set is obtained, and when the current accuracy is greater than the preset accuracy, the pre-trained language model after training is marked as the pre-trained language model that has passed the test.

[0098] S26, obtaining current news of the financial platform, inputting the current news into the pre-trained language model that has passed the test, and obtaining a current recognition result output by the pre-trained language model that has passed the test based on the current news.

[0099] Wherein, after obtaining the current news of the financial platform, inputting the current news into the pre-trained language model that has passed the test, and obtaining the current recognition result output by the pre-trained language model that has passed the test based on the current news, the news recognition method includes:

[0100] Obtain a prompt box of the financial platform, and display the current recognition result through the prompt box.

[0101] In the embodiment of the present invention, the beneficial effects are in two aspects. On the one hand, the current news of the financial platform is identified, the current news is input into the pre-trained language model that has passed the test, and the pre-trained language model that has passed the recognition test outputs a current recognition result based on the current news. Since no manual recognition is required, the recognition time of the current news of the financial platform is reduced, which is beneficial to improving the recognition efficiency of the current news of the financial platform. On the other hand, since the pre-trained language model will not be affected by manual intervention, it is beneficial to improve the reliability of the current recognition result.

[0102] See also Figure 3 , Figure 3 yes Figure 1 A specific implementation flow diagram of step S23 is described in detail as follows:

[0103] S31, using a linear transformation layer of the pre-trained language model, converting the first feature vector into a first output vector, and converting the second feature vector into a second output vector;

[0104] S32: Fusing the first output vector and the second output vector through a multi-head self-attention mechanism to generate a third feature vector.

[0105] In the embodiment of the present invention, the third eigenvector includes information of the first eigenvector and the second eigenvector, thereby achieving effective integration and refinement of information.

[0106] See also Figure 4 , Figure 4 yes Figure 1 A specific implementation flow diagram of step S25 is described in detail as follows:

[0107] S41, when the loss value is less than a preset value, obtaining a test set corresponding to the financial platform, using the test set to test the pre-trained language model, and generating a test result;

[0108] S42: According to the test result, obtain the current accuracy of the pre-trained language model on the test set, and when the current accuracy is greater than a preset accuracy, obtain the pre-trained language model that passes the test.

[0109] In an embodiment of the present invention, when the current accuracy rate is greater than the preset accuracy rate, it means that the pre-trained language model performs well on the preset test data set, meets the expected behavior and performance indicators, and enhances the reliability of the pre-trained language model.

[0110] In an embodiment of the present invention, the first embedding vector is reconstructed according to the attention score to generate a second embedding vector corresponding to the semantic unit, which can focus more on important semantic units and improve the accuracy and efficiency of semantic unit screening.

[0111] See also Figure 5 , Figure 5 yes Figure 1 A specific implementation flow diagram of step S26 is described in detail as follows:

[0112] S51, obtaining a chat interface of the financial platform, obtaining a dialogue window of the chat interface, and obtaining the current news appearing in the dialogue window;

[0113] S52, obtaining the news text, news image, and news audio of the current news, performing feature extraction on the news text, news image, and news audio of the current news to obtain a fourth feature vector, inputting the fourth feature vector into the pre-trained language model that has passed the test, and obtaining the current recognition result output by the pre-trained language model that has passed the test based on the fourth feature vector.

[0114] Exemplarily, the news text of the current news includes the title, author, release date, news body, and keywords of the current news.

[0115] In the embodiment of the present invention, since manual recognition is not required, the recognition time of the current news of the financial platform is reduced, which is conducive to improving the recognition efficiency of the current news of the financial platform.

[0116] See also Figure 6 , Figure 6 is a structural diagram of a news identification device in one embodiment of the present invention. Figure 6 As shown, the news identification device includes a first acquisition module 101, an extraction module 102, a generation module 103, a second acquisition module 104, a third acquisition module 105, and an identification module 106. The functional modules are described in detail as follows:

[0117] The first acquisition module 101 is used to acquire confirmed false news and real news on the financial platform, select the false news or the real news as preset news, acquire preset recognition results corresponding to the preset news, and acquire financial expertise corresponding to the financial platform;

[0118] An extraction module 102 is used to extract features from the preset news to obtain a first feature vector, and to extract features from the financial expertise to obtain a second feature vector;

[0119] A generating module 103, configured to fuse the first feature vector and the second feature vector to generate a third feature vector;

[0120] A second acquisition module 104 is used to obtain a predicted recognition result output by a pre-trained language model based on the third feature vector, and generate a loss value between the predicted recognition result and the preset recognition result based on a predefined generation method;

[0121] A third acquisition module 105 is used to acquire the pre-trained language model that passes the test by adopting a predefined acquisition method when the loss value is less than a preset value;

[0122] The recognition module 106 is used to obtain the current news of the financial platform, input the current news into the pre-trained language model that has passed the test, and obtain the current recognition result output by the pre-trained language model that has passed the test based on the current news.

[0123] In one embodiment, the extraction module 102 includes:

[0124] A first extraction subunit is used to obtain news text, news image, and news audio of the preset news, and perform feature extraction on the news text, news image, and news audio of the preset news to obtain a first feature vector;

[0125] The second extraction subunit is used to obtain financial terms and financial regulations in the financial expertise, perform feature extraction on the financial terms and financial regulations, and obtain a second feature vector.

[0126] In one embodiment, the generating module 103 includes:

[0127] a conversion subunit, configured to convert the first feature vector into a first output vector and convert the second feature vector into a second output vector by using a linear transformation layer of the pre-trained language model;

[0128] The first generating subunit is used to fuse the first output vector and the second output vector through a multi-head self-attention mechanism to generate a third feature vector.

[0129] In one embodiment, the second acquisition module 104 includes:

[0130] An output subunit, used to obtain a predicted recognition result output by a classifier of a pre-trained language model based on the third feature vector;

[0131] The second generating subunit is used to generate a loss value between the predicted recognition result and the preset recognition result through a preset cross entropy loss function.

[0132] In one embodiment, the third acquisition module 105 includes:

[0133] A third generating subunit is configured to, when the loss value is less than a preset value, obtain a test set corresponding to the financial platform, use the test set to test the pre-trained language model, and generate a test result;

[0134] The first acquisition subunit is used to acquire the current accuracy of the pre-trained language model on the test set according to the test result, and when the current accuracy is greater than a preset accuracy, acquire the pre-trained language model that passes the test.

[0135] In one embodiment, the identification module 106 includes:

[0136] A second acquisition subunit is used to acquire a chat interface of the financial platform, acquire a dialogue window of the chat interface, and acquire the current news appearing in the dialogue window;

[0137] The recognition subunit is used to obtain the news text, news image, and news audio of the current news, perform feature extraction on the news text, news image, and news audio of the current news to obtain a fourth feature vector, input the fourth feature vector into the pre-trained language model that has passed the test, and obtain the current recognition result output by the pre-trained language model that has passed the test based on the fourth feature vector.

[0138] In one embodiment, the news identification device further includes:

[0139] The display module is used to obtain a prompt box of the financial platform and display the current recognition result through the prompt box.

[0140] In the embodiment of the present invention, the beneficial effects are in two aspects. On the one hand, the current news of the financial platform is identified, the current news is input into the pre-trained language model that has passed the test, and the pre-trained language model that has passed the recognition test outputs a current recognition result based on the current news. Since no manual recognition is required, the recognition time of the current news of the financial platform is reduced, which is beneficial to improving the recognition efficiency of the current news of the financial platform. On the other hand, since the pre-trained language model will not be affected by manual intervention, it is beneficial to improve the reliability of the current recognition result.

[0141] The specific definition of the news identification device can be found in the above-mentioned definition of the news identification method, which will not be repeated here.

[0142] Each module in the above-mentioned news identification device can be implemented in whole or in part by software, hardware or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each module.

[0143] See also Figure 7 , Figure 7 is a schematic diagram of a computer device in one embodiment of the present invention. In one embodiment, a computer device is provided. The computer device may be a server. The internal structure diagram thereof may be as shown in FIG. Figure 7 The computer device includes a processor, a memory, a network interface and a database connected through a system bus.

[0144] The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes non-volatile and / or volatile storage media and internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external client through a network connection. When the computer program is executed by the processor, the functions or steps of a news identification method can be implemented on the server side.

[0145] See also Figure 8 , Figure 8 is another structural diagram of a computer device in one embodiment of the present invention. In one embodiment, a computer device is provided. The computer device may be a client, and its internal structure diagram may be as shown in FIG. Figure 8 As shown. The computer device includes a processor, a memory, a network interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external task database through a network connection. When the computer program is executed by the processor, the functions or steps of a news identification method can be implemented on the client.

[0146] In one embodiment, a computer device is provided, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor.

[0147] It should be noted that the above functions or steps that can be implemented by the computer-readable storage medium or computer device can refer to the relevant descriptions of the server and the client in the aforementioned method embodiment. To avoid repetition, they will not be described one by one here.

[0148] The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a graphics processing unit (GPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components.

[0149] The above description and accompanying drawings fully illustrate the embodiments of the present disclosure so that those skilled in the art can practice them. Other embodiments may include structural, logical, electrical, process and other changes. The embodiments represent possible changes only. Unless explicitly required, separate components and functions are optional, and the order of operation may vary. Parts and subsamples of some embodiments may be included in or replace parts and subsamples of other embodiments. Moreover, the words used in this application are only used to describe the embodiments and are not used to limit the claims. As used in the description of the embodiments and claims, unless the context clearly indicates, the singular forms of "a", "an" and "the" are intended to include plural forms as well. Similarly, the term "and / or" as used in this application refers to any and all possible combinations of listings containing one or more associated ones. In addition, when used in the present application, the term "comprise" and its variants "comprises" and / or comprising refer to the presence of a stated sub-sample, whole, step, operation, element, and / or component, but do not exclude the presence or addition of one or more other sub-samples, wholes, steps, operations, elements, components and / or groups of these. In the absence of further restrictions, the elements defined by the sentence "comprising a ..." do not exclude the presence of other identical elements in the process, method or device comprising the elements. In this article, each embodiment may focus on the differences from other embodiments, and the same and similar parts between the various embodiments may refer to each other. For the methods, products, etc. disclosed in the embodiments, if they correspond to the method part disclosed in the embodiments, then the relevant parts can refer to the description of the method part.

[0150] Those skilled in the art will appreciate that the units 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 may depend on the specific application and design constraints of the technical solution. Technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the embodiments of the present disclosure. Technicians can clearly understand that for the convenience and simplicity of description, the specific working process of the above-described system, device and unit can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0151] In the embodiments disclosed herein, the disclosed methods and products (including but not limited to devices, equipment, etc.) can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of units can be only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some sub-samples can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between each other shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms. The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to implement this embodiment. In addition, each functional unit in the embodiment of the present disclosure may be integrated in a processing unit, or each unit may exist physically alone, or two or more units may be integrated in one unit.

[0152] The flowchart and block diagram in the accompanying drawings show the possible architecture, function and operation of the system, method and computer program product according to the embodiment of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of the code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. In some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, which can depend on the functions involved. In the description corresponding to the flowchart and the block diagram in the accompanying drawings, the operations or steps corresponding to different boxes can also occur in a different order from the order disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, which can depend on the functions involved. Each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented by a dedicated hardware-based system that performs the specified functions or actions, or may be implemented by a combination of dedicated hardware and computer instructions.

Claims

1. A news identification method, characterized in that: include: Acquire confirmed false news and real news on the financial platform, select the false news or the real news as preset news, obtain preset recognition results corresponding to the preset news, and obtain financial expertise corresponding to the financial platform; Extracting features from the preset news to obtain a first feature vector, and extracting features from the financial expertise to obtain a second feature vector; Merging the first eigenvector and the second eigenvector to generate a third eigenvector; Obtaining a predicted recognition result output by a pre-trained language model based on the third feature vector, and generating a loss value between the predicted recognition result and the preset recognition result based on a predefined generation method; When the loss value is less than a preset value, a predefined acquisition method is used to acquire the pre-trained language model that has passed the test; The current news of the financial platform is obtained, the current news is input into the pre-trained language model that has passed the test, and the current recognition result output by the pre-trained language model that has passed the test based on the current news is obtained.

2. The news identification method according to claim 1, characterized in that: The step of extracting features from the preset news to obtain a first feature vector and extracting features from the financial expertise to obtain a second feature vector includes: Acquire news text, news image, and news audio of the preset news, perform feature extraction on the news text, news image, and news audio of the preset news, and obtain a first feature vector; Financial terms and financial regulations in the financial expertise are obtained, and features of the financial terms and financial regulations are extracted to obtain a second feature vector.

3. The news identification method according to claim 1, characterized in that: The step of fusing the first feature vector and the second feature vector to generate a third feature vector includes: Using a linear transformation layer of the pre-trained language model, convert the first feature vector into a first output vector, and convert the second feature vector into a second output vector; The first output vector and the second output vector are fused through a multi-head self-attention mechanism to generate a third feature vector.

4. The news identification method according to claim 1, characterized in that: The obtaining of the predicted recognition result output by the pre-trained language model based on the third feature vector, and generating a loss value between the predicted recognition result and the preset recognition result based on a predefined generation method, includes: Obtaining a predicted recognition result output by a classifier of a pre-trained language model based on the third feature vector; A loss value between the predicted recognition result and the preset recognition result is generated through a preset cross entropy loss function.

5. The news identification method according to claim 1, characterized in that: When the loss value is less than a preset value, a predefined acquisition method is used to acquire the pre-trained language model that has passed the test, including: When the loss value is less than a preset value, obtaining a test set corresponding to the financial platform, using the test set to test the pre-trained language model, and generating a test result; According to the test result, the current accuracy of the pre-trained language model on the test set is obtained, and when the current accuracy is greater than a preset accuracy, the pre-trained language model that passes the test is obtained.

6. The news identification method according to claim 1, characterized in that: The obtaining of the current news of the financial platform, inputting the current news into the pre-trained language model that has passed the test, and obtaining the current recognition result output by the pre-trained language model that has passed the test based on the current news, includes: Acquire a chat interface of the financial platform, acquire a dialogue window of the chat interface, and acquire the current news appearing in the dialogue window; Acquire news text, news image, and news audio of the current news, perform feature extraction on the news text, news image, and news audio of the current news to obtain a fourth feature vector, input the fourth feature vector into the pre-trained language model that has passed the test, and obtain a current recognition result output by the pre-trained language model that has passed the test based on the fourth feature vector.

7. The news identification method according to claim 1, characterized in that: After obtaining the current news of the financial platform, inputting the current news into the pre-trained language model that has passed the test, and obtaining the current recognition result output by the pre-trained language model that has passed the test based on the current news, the news recognition method includes: Obtain a prompt box of the financial platform, and display the current recognition result through the prompt box.

8. A news identification device, characterized in that: include: The first acquisition module is used to acquire confirmed false news and real news on the financial platform, select the false news or the real news as preset news, acquire preset recognition results corresponding to the preset news, and acquire financial expertise corresponding to the financial platform; An extraction module, used for performing feature extraction on the preset news to obtain a first feature vector, and performing feature extraction on the financial expertise to obtain a second feature vector; A generating module, used for fusing the first feature vector and the second feature vector to generate a third feature vector; A second acquisition module is used to obtain a predicted recognition result output by a pre-trained language model based on the third feature vector, and generate a loss value between the predicted recognition result and the preset recognition result based on a predefined generation method; A third acquisition module is used to acquire the pre-trained language model that passes the test by adopting a predefined acquisition method when the loss value is less than a preset value; The recognition module is used to obtain the current news of the financial platform, input the current news into the pre-trained language model that has passed the test, and obtain the current recognition result output by the pre-trained language model that has passed the test based on the current news.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the news identification method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the news identification method according to any one of claims 1 to 7 are implemented.

Images