Artificial intelligence-based knowledge deficiency identification method and device, equipment and medium

Abstract

The present application relates to the technical field of artificial intelligence, and more particularly to a knowledge defect identification method and device based on artificial intelligence, equipment and medium. The method recommends speech for voice service data, recommends expressions for video service data, compares recommended speech with real speech to obtain a first comparison result, compares recommended expressions with real expressions to obtain a second comparison result, maps the first comparison result and the second comparison result to obtain a correction vector, predicts the recommendation degree of each service knowledge according to learning data to obtain a recommendation value of each service knowledge, multiplies the correction coefficient corresponding to the service knowledge and the recommendation value to obtain a corrected recommendation value, determines the service knowledge corresponding to the maximum corrected recommendation value as the target knowledge, effectively determines the service knowledge that the business personnel need to learn, improves the efficiency of business personnel training, and further improves the business ability of the business personnel, so that the artificial business of the financial service platform is more reliable and professional.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a method, apparatus, device, and medium for identifying knowledge defects based on artificial intelligence. Background Technology

[0002] With the rise of artificial intelligence technology, AI models have been widely applied in financial service platforms. These platforms can be insurance systems, banking systems, transaction systems, order systems, etc. They can support functions such as shopping, social networking, interactive games, and resource transfer, and can also have functions such as applying for loans, credit cards, or purchasing insurance and wealth management products.

[0003] In financial service platforms, it is inevitable to provide human services to users. For example, human services include product recommendations, customer service, and after-sales processing. Therefore, it is necessary to train the business skills of business personnel. Existing business skills training methods are usually systematic, such as providing a complete business manual for business personnel to study.

[0004] However, the systematic training approach lacks specificity for business personnel, making it difficult for them to target their weaknesses during training, resulting in poor training effectiveness. Therefore, improving the relevance of business training and thereby enhancing the business capabilities of personnel has become an urgent issue to be addressed. Summary of the Invention

[0005] In view of this, embodiments of the present invention provide a knowledge defect identification method, apparatus, device and medium based on artificial intelligence to solve the problem of poor targeting of business training.

[0006] In a first aspect, embodiments of the present invention provide a knowledge defect identification method based on artificial intelligence, the knowledge defect identification method comprising:

[0007] The video and voice service data of the target personnel are acquired, and the semantic understanding model is used to recommend dialogue based on the voice service data to obtain recommended dialogue.

[0008] An expression recommendation model is used to recommend expressions to the video service data to obtain recommended expressions. The recommended expressions are compared with the real expressions in the voice service data to obtain a first comparison result. The recommended expressions are compared with the real expressions in the video service data to obtain a second comparison result.

[0009] According to the preset mapping table, and by combining the first comparison result and the second comparison result, a correction vector is obtained. The correction vector includes correction parameters corresponding to N business knowledge items, where N is an integer greater than one.

[0010] Based on the learning data of each business knowledge corresponding to the target personnel, the recommendation degree of each business knowledge is predicted using a business knowledge recommendation model to obtain the recommendation value corresponding to each business knowledge.

[0011] For any business knowledge, the correction coefficient and the recommendation value corresponding to the business knowledge are multiplied together to obtain the multiplication result. The multiplication result is determined as the correction recommendation value corresponding to the business knowledge. The business knowledge corresponding to the maximum value among all correction recommendation values ​​is determined as the target knowledge. The target knowledge is used to instruct the target personnel to conduct business learning.

[0012] Secondly, embodiments of the present invention provide a knowledge defect identification device based on artificial intelligence, the knowledge defect identification device comprising:

[0013] The dialogue recommendation module is used to acquire video and voice service data of target personnel, and use a semantic understanding model to recommend dialogues based on the voice service data to obtain recommended dialogues.

[0014] The facial expression recommendation module is used to recommend facial expressions to the video service data using an facial expression recommendation model, obtain recommended facial expressions, compare the recommended dialogue with the real dialogue in the voice service data to obtain a first comparison result, and compare the recommended facial expressions with the real facial expressions in the video service data to obtain a second comparison result.

[0015] The parameter mapping module is used to obtain a correction vector by mapping the first comparison result and the second comparison result according to a preset mapping table. The correction vector includes N correction parameters corresponding to business knowledge, where N is an integer greater than one.

[0016] The knowledge recommendation module is used to predict the recommendation degree of each business knowledge based on the learning data of the target personnel for each business knowledge obtained, and to obtain the recommendation value corresponding to each business knowledge.

[0017] The knowledge determination module is used to multiply the correction coefficient and the recommendation value corresponding to any business knowledge, obtain the multiplication result, determine the multiplication result as the correction recommendation value corresponding to the business knowledge, and determine the business knowledge corresponding to the maximum value among all correction recommendation values ​​as the target knowledge. The target knowledge is used to instruct the target personnel to conduct business learning.

[0018] Thirdly, embodiments of the present invention provide a computer device, the computer device including a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the knowledge defect identification method as described in the first aspect.

[0019] Fourthly, embodiments of the present invention provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the knowledge defect identification method as described in the first aspect.

[0020] The beneficial effects of the embodiments of the present invention compared with the prior art are as follows:

[0021] The process involves acquiring video and voice business data from target personnel. A semantic understanding model is used to recommend dialogue from the voice data, and an expression recommendation model is used to recommend expressions from the video data. The recommended dialogue is compared with real dialogue in the voice data to obtain a first comparison result. The recommended expressions are then compared with real expressions in the video data to obtain a second comparison result. Based on a pre-defined mapping table, and by combining the first and second comparison results, a correction vector is generated. This correction vector includes correction parameters corresponding to N business knowledge points. Based on the acquired learning data for each business knowledge point from the target personnel, the business knowledge... The recommendation model predicts the recommendation level for each piece of business knowledge, obtaining a recommendation value for each. For any given piece of business knowledge, the correction coefficient and the recommendation value are multiplied together to obtain the corrected recommendation value. The business knowledge corresponding to the maximum corrected recommendation value among all corrected recommendation values ​​is identified as the target knowledge. By combining business data from actual business operations with daily business knowledge learning data, the model effectively identifies the business knowledge that business personnel need to learn in a targeted manner, improving the efficiency of business personnel training and thus enhancing their business capabilities. This makes the human-based business operations of the financial service platform more reliable and professional. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a schematic diagram of an application environment for a knowledge defect identification method based on artificial intelligence provided in Embodiment 1 of the present invention;

[0024] Figure 2 This is a flowchart illustrating a knowledge defect identification method based on artificial intelligence provided in Embodiment 1 of the present invention;

[0025] Figure 3This is a schematic diagram of the structure of a knowledge defect identification device based on artificial intelligence provided in Embodiment 2 of the present invention;

[0026] Figure 4 This is a schematic diagram of the structure of a computer device provided in Embodiment 3 of the present invention. Detailed Implementation

[0027] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of the invention. However, those skilled in the art will understand that the invention can be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods are omitted so as not to obscure the description of the invention with unnecessary detail.

[0028] It should be understood that, when used in this specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0029] It should also be understood that the term “and / or” as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0030] As used in this specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if [described condition or event] is detected" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once [described condition or event] is detected," or "in response to detection of [described condition or event]."

[0031] Furthermore, in the description of this invention and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0032] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of the invention include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0033] The embodiments of this invention can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that utilize digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0034] Foundational technologies for artificial intelligence generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies mainly encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0035] It should be understood that the sequence number of each step in the following embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0036] To illustrate the technical solution of the present invention, specific embodiments are described below.

[0037] The knowledge defect identification method based on artificial intelligence provided in Embodiment 1 of this invention can be applied to, for example... Figure 1In this application environment, the client communicates with the server. The client includes, but is not limited to, handheld computers, desktop computers, laptops, ultra-mobile personal computers (UMPCs), netbooks, cloud terminal devices, and personal digital assistants (PDAs). The server can be a standalone server or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0038] The client and server can be deployed within a financial service platform, which can be an insurance system, banking system, transaction system, order system, etc. The financial service platform can support functions such as shopping, social networking, interactive games, and resource transfer. It can also have functions such as applying for loans, credit cards, or purchasing insurance and wealth management products. The server performs a knowledge defect identification task to analyze the business data of business personnel to identify their business knowledge defects, guide them to learn business knowledge, and thus ensure that the financial service platform maintains a high level of human service.

[0039] See Figure 2 This is a flowchart illustrating a knowledge defect identification method based on artificial intelligence provided in Embodiment 1 of the present invention. The aforementioned knowledge defect identification method can be applied to... Figure 1 The server-side computer connects to the client to retrieve video and audio business data of the target personnel. This data can be obtained by recording video and audio during the target personnel's business activities. The server-side computer has storage capabilities, storing learning data for each business skill corresponding to the target personnel. For example... Figure 2 As shown, the knowledge deficiency identification method may include the following steps:

[0040] Step S201: Obtain video and voice service data of the target personnel, and use a semantic understanding model to recommend dialogue based on the voice service data to obtain recommended dialogue.

[0041] Among them, the target personnel can refer to the business personnel in the financial service platform who provide manual service projects and need to undergo business knowledge training; the video business data can refer to the video recordings obtained by the target personnel when they conduct business, and the video recording equipment can refer to video recorders, cameras, handheld photography equipment, etc.; the voice business data can refer to the audio recordings obtained by the target personnel when they conduct business, and the audio recording equipment can refer to voice recorders, microphones, monitoring headphones, etc.

[0042] Semantic understanding models can be used to extract semantic information contained in voice service data and recommend dialogue based on the extracted semantic information. Recommended dialogue can refer to the dialogue to be used in the semantic scenario contained in the voice service data. Dialogue can refer to the communication method and content.

[0043] Optionally, the semantic understanding model includes a text converter, an encoder, and a decoder;

[0044] Using a semantic understanding model, speech recommendations are performed on voice service data to obtain recommended speech, including:

[0045] A text converter is used to convert voice service data into text service data, and an encoder is used to extract semantic information from the text service data to obtain semantic features.

[0046] Based on semantic features, a decoder is used to reconstruct the text, and the reconstructed text is determined to be the recommended script.

[0047] Among them, the text converter can be used to convert the input voice service data into text service data, the encoder can be used to extract the semantic features of the text service data, the decoder can be used to reconstruct the semantic features into recommended dialogue, the text service data can refer to the text representation of the voice service data, and the text reconstruction result can refer to the output result of the decoder performing the text reconstruction task.

[0048] Specifically, text converters can be implemented using speech recognition models, such as the Seq2Seq model, the CTC model, and the Transducer model.

[0049] In this embodiment, a text converter is used to convert voice service data into text service data, thereby realizing the semantic feature extraction and reconstruction process from text to text. This reduces the difficulty of the text reconstruction task and improves the accuracy of the speech recommendation.

[0050] Optionally, based on semantic features, a decoder is used to reconstruct the text, resulting in the following text reconstruction:

[0051] The semantic features are compared with at least two recommended features in the preset feature set to calculate the similarity, and the first similarity of the corresponding recommended features is obtained. The recommended feature corresponding to the maximum value among all first similarities is determined as the feature to be reconstructed.

[0052] The decoder is used to reconstruct the text based on the features to be reconstructed, and the text reconstruction result is obtained.

[0053] The preset feature set includes at least two recommended features. The recommended features can refer to the semantic features corresponding to the preset reference utterances. The first similarity can be used to represent the degree of difference between the semantic features and the corresponding recommended features. The features to be reconstructed can refer to the recommended features used for text reconstruction.

[0054] Specifically, similarity calculation can be performed using distance metrics such as Euclidean distance, cosine similarity, and Manhattan distance. In this embodiment, cosine similarity is used for similarity calculation. The value range of cosine similarity is [-1, 1]. Correspondingly, the closer the first similarity is to 1, the smaller the difference between the semantic feature and the corresponding recommended feature, and the more similar the semantic feature and the corresponding recommended feature are. The closer the first similarity is to -1, the greater the difference between the semantic feature and the corresponding recommended feature, and the less similar the semantic feature and the corresponding recommended feature are.

[0055] Calculate the similarity between each recommended feature and the semantic feature in the preset feature set to obtain the first similarity corresponding to each recommended feature. Take the maximum value of all first similarities and take the recommended feature corresponding to the maximum value as the feature to be reconstructed. If the maximum value corresponds to multiple recommended features, then select one of all recommended features corresponding to the maximum value as the feature to be reconstructed.

[0056] In this embodiment, by setting a preset feature set, the input of the decoder, i.e. the features to be reconstructed, is selected from the recommended features contained in the preset feature set. Using the preset recommended features can ensure the accuracy of the text reconstruction result. On the other hand, it allows the decoder's text reconstruction task to use the prior recommended features as input. The implementer can directly store the text reconstruction result corresponding to the recommended features as a mapping table, and then directly obtain the text reconstruction result by looking up the mapping table, without using the decoder to perform inference for the text reconstruction task, thereby improving the efficiency and accuracy of the speech recommendation process.

[0057] The steps described above—acquiring video and voice business data of target personnel, using a semantic understanding model to recommend scripts from the voice business data, and obtaining recommended scripts—provide a reference for subsequent analysis of business knowledge deficiencies in target personnel, thereby effectively improving the accuracy of business knowledge deficiency analysis and ultimately increasing the efficiency of business personnel training.

[0058] Step S202: Use the expression recommendation model to recommend expressions to the video service data to obtain recommended expressions. Compare the recommended expressions with the real expressions in the voice service data to obtain the first comparison result. Compare the recommended expressions with the real expressions in the video service data to obtain the second comparison result.

[0059] Among them, the expression recommendation model can be used to extract scene information of target personnel in video business data and recommend expressions based on the extracted scene information. For example, the expression recommendation model can adopt a deep neural network model, which can include an encoder and a classifier. The encoder extracts features from the scene information, and the classifier maps the extracted feature information to the recommended expressions. The encoder can adopt the encoder structure of models such as SSD, FAST, and YOLO, and the classifier can be implemented using fully connected layers. The recommended expressions can refer to the expressions recommended under the scene information contained in the video business data. Expressions can include happiness, anger, surprise, fear, disgust, and sadness, etc.

[0060] Authentic dialogue refers to the dialogue actually used by the target personnel in voice service data, and authentic facial expressions refer to the facial expressions actually used by the target personnel in video service data. The first comparison result can be used to characterize the difference information between recommended dialogue and authentic dialogue, and the second comparison result can be used to characterize the difference information between recommended facial expressions and authentic facial expressions.

[0061] Optionally, knowledge defect identification methods also include:

[0062] Based on voice service data and semantic features, a tone recognition model is used to extract tone and obtain the true tone.

[0063] Obtain the recommended tone corresponding to the recommended speech, compare the recommended tone with the real tone, and obtain the third comparison result;

[0064] Accordingly, based on the preset mapping table and combining the first and second alignment results, a correction vector is obtained, including:

[0065] Based on the mapping table, and by combining the first, second, and third alignment results, a correction vector is obtained.

[0066] The tone recognition model can be used to extract tone information contained in voice service data. It should be noted that the tone recognition model needs to be pre-trained. The training data consists of sample service voice and its corresponding sample tone. The sample service voice can refer to historically collected voice service data, and the sample tone can be obtained by spectral analysis of the sample service voice. The tone recognition model predicts the tone of the sample service voice to obtain the predicted tone. The cross-entropy loss is calculated based on the predicted tone and the sample tone. The tone recognition model can then be pre-trained based on the cross-entropy loss.

[0067] The actual tone can refer to the tone used by the target person in actual communication in the voice service data. The recommended tone can refer to the tone recommended to be used under the semantic information contained in the voice service data. One recommended tone corresponds to one recommended speech. The correspondence between the recommended speech and the recommended tone can be stored in the corresponding computer device on the server side. The third comparison result can be used to characterize the difference information between the recommended tone and the actual tone.

[0068] The mapping table can contain the mapping relationship between the first alignment result, the second alignment result, and the third alignment result and the correction vector.

[0069] Specifically, the comparison process can also use distance metrics, such as the Euclidean distance, cosine similarity, Manhattan distance, etc. In this embodiment, cosine similarity is used for data comparison.

[0070] In this embodiment, additional tone information is added to assist in obtaining the correction vector, which makes the subsequent identification of business knowledge defects based on the correction vector more accurate, thereby improving the efficiency of business knowledge learning, that is, improving the efficiency of business training.

[0071] Optionally, an expression recognition model can be used to recommend expressions from video data, including:

[0072] Use an expression recognition model to identify expressions in video service data to obtain realistic expressions;

[0073] The similarity between the real expression and at least two reference expressions in the preset expression set is calculated to obtain the second similarity of the corresponding reference expressions. The reference expression corresponding to the maximum value among all the second similarities is determined as the recommended expression.

[0074] The preset emoji set can include at least two reference emojis. The reference emojis can refer to preset standard emojis. The second similarity can be used to represent the degree of difference between the real emoji and the corresponding reference emoji.

[0075] Specifically, the closer the second similarity is to 1, the smaller the difference between the real expression and the corresponding recommended expression, and the more similar the real expression and the corresponding recommended expression are. The closer the first similarity is to -1, the greater the difference between the real expression and the corresponding recommended expression, and the less similar the real expression and the corresponding recommended expression are.

[0076] It should be noted that, in order to more accurately determine the recommended expressions, both real expressions and reference expressions can be represented in the form of expression vectors. Expression vectors can refer to the expression features extracted by the expression recognition model.

[0077] In this embodiment, business knowledge defects are identified from multiple dimensions of underlying business capability data, such as the target personnel's language logic, tone of voice, and facial expressions. This improves the accuracy of business knowledge defect identification and thus enhances the efficiency of business training.

[0078] The above steps involve using an expression recommendation model to recommend expressions from video business data, comparing the recommended expressions with the actual expressions in the voice business data to obtain a first comparison result, and comparing the recommended expressions with the actual expressions in the video business data to obtain a second comparison result. This process identifies business knowledge deficiencies from multiple dimensions of underlying business capability data, such as the target personnel's language logic and expressions, thereby improving the accuracy of business knowledge deficiency identification and ultimately increasing the efficiency of business training.

[0079] Step S203: Based on the preset mapping table, and by combining the first comparison result and the second comparison result, a correction vector is obtained. The correction vector includes correction parameters corresponding to N business knowledge items.

[0080] Where N is an integer greater than one, the preset mapping table may include the mapping relationship between the first comparison result, the second comparison result and the correction vector, and one correction parameter corresponds to one piece of business knowledge.

[0081] Specifically, the correction vector can be represented as S, then S = {s1, s2, ..., s} N The value of N is determined by the amount of business knowledge, which can include communication skills, product knowledge, etc.

[0082] The above steps, based on a preset mapping table and combining the first and second comparison results, yield a correction vector. This correction vector includes correction parameters corresponding to N business knowledge items. This provides adjustments for subsequent recommendation value correction based on the target personnel's own abilities and qualities, making the corrected recommendation value more consistent with the target personnel's characteristics. This effectively identifies the business knowledge that business personnel need to learn in a targeted manner, thus improving the efficiency of business personnel training.

[0083] Step S204: Based on the learning data of each business knowledge corresponding to the target personnel, use the business knowledge recommendation model to predict the recommendation degree of each business knowledge and obtain the recommendation value corresponding to each business knowledge.

[0084] The learning data can refer to the learning data of target personnel on business knowledge, such as learning duration and learning frequency. The business knowledge recommendation model can predict the recommendation degree of each business knowledge based on the learning data of target personnel for each business knowledge. The recommendation value can refer to the recommended learning degree of the corresponding business knowledge.

[0085] Optionally, the learning data includes average learning duration, learning frequency, and total learning duration;

[0086] Based on the acquired learning data for each business knowledge item for the target personnel, a business knowledge recommendation model is used to predict the recommendation level of each business knowledge item, resulting in a recommendation value for each business knowledge item, including:

[0087] For any given business knowledge, the average learning time, learning frequency, and total learning time of the target personnel for that business knowledge are combined into a learning data vector.

[0088] A business knowledge recommendation model is used to map learned data vectors to recommended values ​​for corresponding business knowledge.

[0089] Among them, average learning time can refer to the average time taken for target personnel to learn a certain business data multiple times, learning frequency can represent the interval between multiple learning sessions for target personnel on a certain business data, and total learning time can refer to the total sum of the time taken for target personnel to learn a certain business data multiple times.

[0090] A learning data vector can be a vector representation of the average learning time, learning frequency, and total learning time of the target personnel for the corresponding business knowledge.

[0091] Specifically, the business knowledge recommendation model can be implemented using a fully connected layer. The number of input neurons in the fully connected layer corresponds to the number of learning data types in the learning data vector, and the number of output neurons in the fully connected layer corresponds to the number of business knowledge types.

[0092] Optionally, a business knowledge recommendation model can be used to map the learned data vectors to recommended values ​​corresponding to business knowledge, including:

[0093] Obtain a reference learning vector, and use a business knowledge recommendation model to perform a difference analysis on the learning data vector and the reference learning vector to obtain the difference analysis results.

[0094] Map the difference analysis results to recommended values.

[0095] The reference learning vector can refer to the learning vector of the reference personnel. The reference personnel can be determined based on user evaluations, internal assessments, etc. The reference personnel can refer to business personnel with excellent business performance. The difference analysis results can characterize the degree of difference between the learning data vector and the reference learning vector.

[0096] Specifically, in this embodiment, the business knowledge recommendation model can adopt a Siamese network structure, which contains two completely identical branches. Each branch contains an encoder, which is used to extract feature information of the input learning vector. The difference analysis result can refer to the degree of difference between the feature information corresponding to the learning data vector and the reference learning vector, respectively.

[0097] The above steps involve using a business knowledge recommendation model to predict the recommendation level of each business knowledge based on the learning data of the target personnel for each business knowledge, and obtaining the recommendation value corresponding to each business knowledge. By comparing the learning data of the target personnel and the reference personnel, the business knowledge deficiency analysis results of the target personnel are obtained. Subsequently, by combining the correction parameters determined based on the characteristics of the target personnel, a reliable corrected business knowledge deficiency analysis result is obtained, which improves the efficiency of business personnel training and thus improves the business capabilities of the business personnel.

[0098] Step S205: For any business knowledge, multiply the correction coefficient and the recommendation value corresponding to the business knowledge to obtain the multiplication result. Determine the multiplication result as the correction recommendation value of the corresponding business knowledge, and determine the business knowledge corresponding to the maximum value among all correction recommendation values ​​as the target knowledge.

[0099] The multiplication result can refer to the calculation result obtained by multiplying the correction coefficient corresponding to the business knowledge and the recommended value. The target knowledge is used to instruct the target personnel to conduct business learning.

[0100] The above steps, which involve multiplying the correction coefficient and recommended value corresponding to any business knowledge to obtain the multiplication result, determining the multiplication result as the corrected recommended value for the corresponding business knowledge, and identifying the business knowledge corresponding to the maximum value among all corrected recommended values ​​as the target knowledge, combine business data from actual business development processes with daily business knowledge learning data to effectively determine the business knowledge that business personnel need to learn in a targeted manner, thereby improving the efficiency of business personnel training.

[0101] In this embodiment, business data from actual business operations and daily business knowledge learning data are combined to effectively identify the business knowledge that business personnel need to learn in a targeted manner, thereby improving the efficiency of business personnel training, enhancing their business capabilities, and making the manual business of the financial service platform more reliable and professional.

[0102] Corresponding to the knowledge defect identification method based on artificial intelligence in the above embodiments, Figure 3 This diagram illustrates the structural block diagram of an AI-based knowledge defect identification device according to Embodiment 2 of the present invention. The device is applied to a server, and the corresponding computer device is connected to a client to obtain video and voice service data of the target personnel from the client. The video and voice service data can be obtained by recording video and audio during the target personnel's business activities. The server-side computer device has storage capabilities, storing learning data for each business knowledge item corresponding to the target personnel. For ease of explanation, only the parts relevant to the embodiments of the present invention are shown.

[0103] See Figure 3 The knowledge defect identification device includes:

[0104] The script recommendation module 31 is used to acquire video and voice service data of the target personnel, and use a semantic understanding model to recommend scripts from the voice service data.

[0105] The expression recommendation module 32 is used to recommend expressions to video service data using an expression recommendation model, obtain recommended expressions, compare the recommended dialogue with the real dialogue in the voice service data to obtain a first comparison result, and compare the recommended expressions with the real expressions in the video service data to obtain a second comparison result.

[0106] The parameter mapping module 33 is used to obtain a correction vector by combining the first comparison result and the second comparison result according to the preset mapping table. The correction vector includes correction parameters corresponding to N business knowledge, where N is an integer greater than one.

[0107] The knowledge recommendation module 34 is used to predict the recommendation degree of each business knowledge based on the learning data of the target personnel for each business knowledge, and to obtain the recommendation value corresponding to each business knowledge.

[0108] The knowledge determination module 35 is used to multiply the correction coefficient and the recommended value corresponding to any business knowledge, obtain the multiplication result, determine the multiplication result as the corrected recommended value of the corresponding business knowledge, and determine the business knowledge corresponding to the maximum value among all corrected recommended values ​​as the target knowledge. The target knowledge is used to instruct the target personnel to carry out business learning.

[0109] Optionally, the semantic understanding model includes a text converter, an encoder, and a decoder;

[0110] The aforementioned script recommendation module 31 includes:

[0111] The semantic extraction unit is used to convert voice service data into text service data using a text converter, and to extract semantic information from the text service data using an encoder to obtain semantic features.

[0112] The text reconstruction unit is used to reconstruct text based on semantic features using a decoder, obtain the text reconstruction result, and determine the text reconstruction result as the recommended wording.

[0113] Optionally, the above text reconstruction unit includes:

[0114] The feature comparison subunit is used to calculate the similarity between the semantic feature and at least two recommended features in the preset feature set, obtain the first similarity of the corresponding recommended feature, and determine the recommended feature corresponding to the maximum value among all first similarities as the feature to be reconstructed.

[0115] The feature decoding subunit is used to reconstruct the text using the decoder to obtain the text reconstruction result.

[0116] Optionally, the aforementioned knowledge defect identification device further includes:

[0117] The pitch extraction module is used to extract pitches based on voice service data and semantic features using a pitch recognition model to obtain the true pitch.

[0118] The pitch comparison module is used to obtain the recommended pitch corresponding to the recommended speech, compare the recommended pitch with the real pitch, and obtain the third comparison result;

[0119] Accordingly, the parameter mapping module 33 includes:

[0120] The vector mapping subunit is used to obtain the correction vector by mapping the first alignment result, the second alignment result, and the third alignment result according to the mapping table.

[0121] Optionally, the aforementioned emoji recommendation module 32 includes:

[0122] The facial expression recognition unit is used to perform facial expression recognition on video service data using an facial expression recognition model to obtain real facial expressions;

[0123] The expression comparison unit is used to calculate the similarity between the real expression and at least two reference expressions in the preset expression set, obtain the second similarity of the corresponding reference expressions, and determine the reference expression corresponding to the maximum value among all the second similarities as the recommended expression.

[0124] Optionally, the learning data includes average learning duration, learning frequency, and total learning duration;

[0125] The aforementioned knowledge recommendation module 34 includes:

[0126] Vector composition unit, used to form a learning data vector by combining the average learning time, learning frequency and total learning time of the target personnel for any business knowledge;

[0127] The vector mapping unit is used to map learning data vectors to corresponding business knowledge recommendation values ​​using a business knowledge recommendation model.

[0128] Optionally, the above vector mapping unit includes:

[0129] The difference analysis subunit is used to obtain the reference learning vector, and to perform difference analysis on the learning data vector and the reference learning vector using the business knowledge recommendation model to obtain the difference analysis results.

[0130] The recommended value mapping subunit is used to map the difference analysis results to recommended values.

[0131] It should be noted that the information interaction and execution process between the above modules, units, and sub-units are based on the same concept as the method embodiments of the present invention. For details on their specific functions and technical effects, please refer to the method embodiments section, which will not be repeated here.

[0132] Figure 4 This is a schematic diagram of the structure of a computer device provided in Embodiment 3 of the present invention. Figure 4 As shown, the computer device of this embodiment includes: at least one processor ( Figure 4 The diagram shows only one of the following: a memory and a computer program stored in the memory and capable of running on at least one processor. When the processor executes the computer program, it implements the steps in any of the above embodiments of the knowledge defect identification method.

[0133] This computer device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that... Figure 4 The examples of computer devices are merely examples and do not constitute a limitation on computer devices. Computer devices may include more or fewer components than shown in the illustration, or combinations of certain components, or different components, such as network interfaces, displays, and input devices.

[0134] The processor referred to can be a CPU, but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0135] Memory includes readable storage media, internal memory, etc., wherein internal memory can be the RAM of a computer device, providing an environment for the operation of the operating system and computer-readable instructions stored in the readable storage media. The readable storage media can be the hard drive of a computer device, or in other embodiments, it can be an external storage device of the computer device, such as a plug-in hard drive, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card. Furthermore, memory can include both internal storage units and external storage devices of a computer device. Memory is used to store the operating system, applications, bootloader, data, and other programs, such as program code for computer programs. Memory can also be used to temporarily store data that has been output or will be output.

[0136] Those skilled in the art will understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the functions described above can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this invention. The specific working process of the units and modules in the above device can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here. If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention can implement all or part of the processes in the methods of the above embodiments by instructing related hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the above method embodiments. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. A computer-readable medium can include at least: any entity or device capable of carrying computer program code, a recording medium, a computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electrical carrier signals or telecommunication signals.

[0137] The present invention can implement all or part of the processes in the methods of the above embodiments, or it can be accomplished by a computer program product. When the computer program product is run on a computer device, the computer device executes the steps in the above method embodiments.

[0138] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0139] Those skilled in the art will recognize that the units and algorithm steps of the various examples 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 implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0140] In the embodiments provided by this invention, it should be understood that the disclosed apparatus / computer devices and methods can be implemented in other ways. For example, the apparatus / computer device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0141] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0142] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A knowledge defect identification method based on artificial intelligence, characterized in that, The knowledge defect identification method includes: The video and voice service data of the target personnel are acquired, and the semantic understanding model is used to recommend dialogue based on the voice service data to obtain recommended dialogue. An expression recommendation model is used to recommend expressions to the video service data to obtain recommended expressions. The recommended expressions are compared with the real expressions in the voice service data to obtain a first comparison result. The recommended expressions are compared with the real expressions in the video service data to obtain a second comparison result. According to the preset mapping table, and by combining the first comparison result and the second comparison result, a correction vector is obtained. The correction vector includes correction parameters corresponding to N business knowledge items, where N is an integer greater than one. Based on the learning data of each business knowledge corresponding to the target personnel, the recommendation degree of each business knowledge is predicted using a business knowledge recommendation model to obtain the recommendation value corresponding to each business knowledge. For any business knowledge, the correction coefficient and the recommendation value corresponding to the business knowledge are multiplied together to obtain the multiplication result. The multiplication result is determined as the correction recommendation value corresponding to the business knowledge. The business knowledge corresponding to the maximum value among all correction recommendation values ​​is determined as the target knowledge. The target knowledge is used to instruct the target personnel to conduct business learning.

2. The knowledge defect identification method according to claim 1, characterized in that, The semantic understanding model includes a text converter, an encoder, and a decoder; The step of using a semantic understanding model to recommend dialogue based on the voice service data, and obtaining recommended dialogue, includes: The text converter is used to convert the voice service data into text service data, and the encoder is used to extract the semantic information of the text service data to obtain semantic features. Based on the semantic features, the decoder is used to reconstruct the text, and the reconstructed text result is determined to be the recommended script.

3. The knowledge defect identification method according to claim 2, characterized in that, The step of reconstructing the text using the decoder based on the semantic features to obtain the text reconstruction result includes: The semantic features are compared with at least two recommended features in the preset feature set to calculate the similarity, and the first similarity of the corresponding recommended features is obtained. The recommended feature corresponding to the maximum value among all first similarities is determined as the feature to be reconstructed. The decoder is used to reconstruct the text of the feature to be reconstructed, and the text reconstruction result is obtained.

4. The knowledge defect identification method according to claim 2, characterized in that, The knowledge defect identification method also includes: Based on the voice service data and the semantic features, a tone recognition model is used to extract the tone and obtain the true tone. Obtain the recommended tone corresponding to the recommended speech, compare the recommended tone with the real tone, and obtain a third comparison result; Accordingly, the step of obtaining the correction vector by mapping the first comparison result and the second comparison result according to the preset mapping table includes: The correction vector is obtained by mapping the first alignment result, the second alignment result, and the third alignment result according to the mapping table.

5. The knowledge defect identification method according to claim 1, characterized in that, The step of using an expression recognition model to recommend expressions from the video service data, and obtaining recommended expressions, includes: The facial expression recognition model is used to perform facial expression recognition on the video service data to obtain the real facial expression; The similarity between the real expression and at least two reference expressions in the preset expression set is calculated to obtain the second similarity of the corresponding reference expressions. The reference expression corresponding to the maximum value among all the second similarities is determined as the recommended expression.

6. The knowledge defect identification method according to any one of claims 1 to 5, characterized in that, The learning data includes average learning time, learning frequency, and total learning time; The step of using a business knowledge recommendation model to predict the recommendation level of each business knowledge based on the acquired learning data of the target personnel for each business knowledge, and obtaining the recommendation value corresponding to each business knowledge, includes: For any business knowledge, a learning data vector is formed by the average learning time, the learning frequency and the total learning time of the target personnel for the corresponding business knowledge. The business knowledge recommendation model is used to map the learning data vector to the corresponding recommendation value of the business knowledge.

7. The knowledge defect identification method according to claim 6, characterized in that, The step of mapping the learning data vector to a recommended value corresponding to the business knowledge using the business knowledge recommendation model includes: Obtain a reference learning vector, and use the business knowledge recommendation model to perform a difference analysis on the learning data vector and the reference learning vector to obtain the difference analysis results; The difference analysis results are mapped to the recommended values.

8. A knowledge defect identification device based on artificial intelligence, characterized in that, The knowledge defect identification device includes: The dialogue recommendation module is used to acquire video and voice service data of target personnel, and use a semantic understanding model to recommend dialogues based on the voice service data to obtain recommended dialogues. The facial expression recommendation module is used to recommend facial expressions to the video service data using an facial expression recommendation model, obtain recommended facial expressions, compare the recommended dialogue with the real dialogue in the voice service data to obtain a first comparison result, and compare the recommended facial expressions with the real facial expressions in the video service data to obtain a second comparison result. The parameter mapping module is used to obtain a correction vector by mapping the first comparison result and the second comparison result according to a preset mapping table. The correction vector includes N correction parameters corresponding to business knowledge, where N is an integer greater than one. The knowledge recommendation module is used to predict the recommendation degree of each business knowledge based on the learning data of the target personnel for each business knowledge obtained, and to obtain the recommendation value corresponding to each business knowledge. The knowledge determination module is used to multiply the correction coefficient and the recommendation value corresponding to any business knowledge, obtain the multiplication result, determine the multiplication result as the correction recommendation value corresponding to the business knowledge, and determine the business knowledge corresponding to the maximum value among all correction recommendation values ​​as the target knowledge. The target knowledge is used to instruct the target personnel to conduct business learning.

9. A computer device, characterized in that, The computer device includes a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the knowledge defect identification method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the knowledge defect identification method as described in any one of claims 1 to 7.

Images