CN116308735A - Financial data prediction method, device, electronic equipment and storage medium - Google Patents
Financial data prediction method, device, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN116308735A CN116308735A CN202310093442.4A CN202310093442A CN116308735A CN 116308735 A CN116308735 A CN 116308735A CN 202310093442 A CN202310093442 A CN 202310093442A CN 116308735 A CN116308735 A CN 116308735A
- Authority
- CN
- China
- Prior art keywords
- consumption preference
- target user
- user
- financial data
- consumption
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 74
- 238000012545 processing Methods 0.000 claims abstract description 8
- 238000012549 training Methods 0.000 claims description 57
- 238000013145 classification model Methods 0.000 claims description 36
- 238000006243 chemical reaction Methods 0.000 claims description 15
- 238000004590 computer program Methods 0.000 claims description 10
- 238000010276 construction Methods 0.000 claims description 9
- 238000004458 analytical method Methods 0.000 claims description 8
- 238000010586 diagram Methods 0.000 description 8
- 238000005516 engineering process Methods 0.000 description 8
- 238000004891 communication Methods 0.000 description 7
- 238000010801 machine learning Methods 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- 230000004927 fusion Effects 0.000 description 4
- 230000006399 behavior Effects 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000007246 mechanism Effects 0.000 description 3
- 230000007547 defect Effects 0.000 description 2
- 230000001419 dependent effect Effects 0.000 description 2
- 230000008451 emotion Effects 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000003058 natural language processing Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- FGUUSXIOTUKUDN-IBGZPJMESA-N C1(=CC=CC=C1)N1C2=C(NC([C@H](C1)NC=1OC(=NN=1)C1=CC=CC=C1)=O)C=CC=C2 Chemical compound C1(=CC=CC=C1)N1C2=C(NC([C@H](C1)NC=1OC(=NN=1)C1=CC=CC=C1)=O)C=CC=C2 FGUUSXIOTUKUDN-IBGZPJMESA-N 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 239000000654 additive Substances 0.000 description 1
- 230000000996 additive effect Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000002457 bidirectional effect Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000001556 precipitation Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000013077 scoring method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000013526 transfer learning Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
- 230000003442 weekly effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/06—Creation of reference templates; Training of speech recognition systems, e.g. adaptation to the characteristics of the speaker's voice
- G10L15/063—Training
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
- G10L15/18—Speech classification or search using natural language modelling
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/26—Speech to text systems
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L17/00—Speaker identification or verification techniques
- G10L17/04—Training, enrolment or model building
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/12—Protocols specially adapted for proprietary or special-purpose networking environments, e.g. medical networks, sensor networks, networks in vehicles or remote metering networks
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Multimedia (AREA)
- Acoustics & Sound (AREA)
- Human Computer Interaction (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Theoretical Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Molecular Biology (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Medical Informatics (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biophysics (AREA)
- Databases & Information Systems (AREA)
- Biomedical Technology (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention provides a financial data prediction method, a device, electronic equipment and a storage medium, and relates to the technical field of data processing, wherein the method comprises the following steps: acquiring dialogue voice of a target user and user information of the target user; the dialogue speech is generated by the target user in a financial dialogue scene; classifying the consumption preference of the target user according to the dialogue voice to obtain the consumption preference level of the target user; acquiring a financial data prediction result of the target user according to the consumption preference level and the user information; the financial data prediction results comprise a payoff amount prediction result and/or a consumption level prediction result. The invention improves the comprehensiveness, objectivity and accuracy of the financial data prediction of the target user and effectively improves the prediction efficiency of the financial data.
Description
Technical Field
The present invention relates to the field of data processing technologies, and in particular, to a method and apparatus for predicting financial data, an electronic device, and a storage medium.
Background
With the development of social economy, the income level of people is continuously improved, and the consumption capability is also continuously improved, and the improvement is not only reflected in the consumption quantity, but also in the consumption quality. When the bank carries out consumption and paying money to the individual customer, the risk and the income must be considered simultaneously, namely, when the income of the bank is improved through the individual consumption and paying money, the risk brought by the infringement of the individual customer must also be prevented, if the paying money is too little, the bank is not beneficial to improving the self profit level, and if the paying money is too much, the client can be caused to generate the infringement action, so that the credit risk is increased. Therefore, how to accurately predict the financial data is one of the technical problems to be solved in the target industry.
At present, when a bank performs consumption and paying for an individual customer, in the first stage, the experience of a customer manager is mainly relied on to judge how much money should be paid for the individual customer, and the method is excessively dependent on subjective judgment of the user, so that the prediction result is inaccurate.
In order to solve the above problems, a second stage of financial data prediction method is proposed, in which a metering economy model is built by means of personal information of a personal client after the personal information is filled in by the personal client, so as to objectively determine how much money should be paid to the consumer, but this way is excessively dependent on the personal information of the client, and the factors affecting the amount of money to be paid are more. Thus, relying solely on personal user information does not accurately predict funds for a payout.
Disclosure of Invention
The invention provides a financial data prediction method, a device, electronic equipment and a storage medium, which are used for solving the defect that in the prior art, financial data cannot be accurately predicted depending on subjective judgment of a user or objective judgment based on personal user information, and improving the prediction accuracy of the financial data.
The invention provides a financial data prediction method, which comprises the following steps:
acquiring dialogue voice of a target user and user information of the target user; the dialogue speech is generated by the target user in a financial dialogue scene;
classifying the consumption preference of the target user according to the dialogue voice to obtain the consumption preference level of the target user;
acquiring a financial data prediction result of the target user according to the consumption preference level and the user information; the financial data prediction results comprise a payoff amount prediction result and/or a consumption level prediction result.
According to the financial data prediction method provided by the invention, the consumption preference of the target user is classified according to the dialogue voice to obtain the consumption preference level of the target user, and the method comprises the following steps:
inputting the dialogue voice into a voice conversion model, and converting the dialogue voice from voice data into text data to obtain text information corresponding to the dialogue voice;
And inputting the text information into a classification model, and classifying the consumption preference of the target user to obtain the consumption preference grade of the target user.
According to the financial data prediction method provided by the invention, the classification model is trained based on the following steps:
acquiring text information corresponding to dialogue voice of a sample user and a consumption preference grade label of the sample user;
constructing a training data set according to the text information corresponding to the sample user and the consumption preference grade label of the sample user;
based on the training data set, fine tuning is carried out on parameters of a pre-training model to obtain the classification model; the pre-training model is generated based on BERT model construction.
According to the financial data prediction method provided by the invention, the consumption preference grade label of the sample user is obtained, and the method comprises the following steps:
acquiring a plurality of initial consumption preference grade labels of the sample user; the plurality of initial consumption preference grade labels are generated by marking the consumption preference grade of the sample user through a plurality of marking objects and/or a plurality of marking rules;
and fusing the multiple initial consumption preference grade labels of the sample user to obtain the consumption preference grade label of the sample user.
According to the financial data prediction method provided by the invention, a training data set is constructed according to text information corresponding to the sample user and consumption preference grade labels of the sample user, and the method comprises the following steps:
carrying out data enhancement on the text information corresponding to the sample user;
constructing the training data set according to the enhanced text information and the consumption preference grade label of the sample user;
wherein the data enhancement includes modifying the target vocabulary.
According to the financial data prediction method provided by the invention, after classifying the consumption preference of the target user according to the dialogue voice to obtain the consumption preference level of the target user, the method further comprises:
and expanding the training data set based on the text information corresponding to the target user and the consumption preference level of the target user.
According to the financial data prediction method provided by the invention, the financial data of the target user is predicted according to the consumption preference level and the user information, and the method comprises the following steps:
inputting the consumption preference level and the user information into a financial data prediction model, and carrying out attribution analysis and financial data prediction on the consumption preference level and the user information to obtain a financial data prediction result of the target user;
The financial data prediction model may be generated based on a regression model and a tree model construction.
The invention also provides a financial data prediction device, which comprises:
the system comprises an acquisition module, a processing module and a processing module, wherein the acquisition module is used for acquiring dialogue voice of a target user and user information of the target user; the dialogue speech is generated by the target user in a financial dialogue scene;
the classification module is used for classifying the consumption preference of the target user according to the dialogue voice to obtain the consumption preference grade of the target user;
the prediction module is used for acquiring a financial data prediction result of the target user according to the consumption preference level and the user information; the financial data prediction results comprise a payoff amount prediction result and/or a consumption level prediction result.
The invention also provides an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the financial data prediction method as described in any one of the above when executing the program.
The present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements a financial data prediction method as described in any of the above.
The invention also provides a computer program product comprising a computer program which when executed by a processor implements a method of predicting financial data as described in any one of the above.
According to the financial data prediction method, the financial data prediction device, the electronic equipment and the storage medium, the consumption preference of the target user is automatically and objectively classified according to the dialogue voice, so that the influence of manual experience is avoided, and the accuracy of the consumption preference classification of the target user is effectively improved; and on the basis of user information, the consumption preference level of the target user is combined, so that the financial data prediction result of the target user is automatically and objectively obtained through various factors related to the financial data prediction result, and the prediction efficiency of the financial data is effectively improved while the comprehensiveness, objectivity and accuracy of the financial data prediction of the target user are improved.
Drawings
In order to more clearly illustrate the invention or the technical solutions of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic flow chart of a financial data prediction method according to the present invention;
FIG. 2 is a schematic diagram of a speech conversion model in a financial data prediction method according to the present invention;
FIG. 3 is a schematic diagram of the BERT model in the financial data prediction method provided by the invention;
FIG. 4 is a second flow chart of the method for predicting financial data according to the present invention;
FIG. 5 is a third flow chart of the financial data prediction method according to the present invention;
FIG. 6 is a schematic diagram of a financial data prediction apparatus according to the present invention;
fig. 7 is a schematic structural diagram of an electronic device provided by the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is apparent that the described embodiments are some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In the past, when paying for individual consumers, the method mainly comprises the following stages:
In the first stage, whether the personal consumer should be paid is judged subjectively through the dialogue between the client manager and the personal consumer, and if the personal consumer can pay the amount of money, the method is very subjective, because subjective preference evaluations of different client managers on the same client are different.
In the second stage, the relevant scholars begin to use a data analysis method to carry out structural modeling, and judge how much to pay to the personal consumer according to personal education background, personal income, personal family condition and whether the personal consumer breaks down in the past, the method is objective, but excessively depends on personal information of the client, and ignores other factors influencing the paying amount, such as neglecting emotion factors when the client manager communicates with the personal consumer, the speaking mode and emotion can be taken as the basis for judgment and cannot be ignored, so that paying funds cannot be accurately predicted only depending on personal user information.
In the third stage, people begin to model objectively by using a combination of subjective and objective modes, but modeling factors not only comprise traditional factors, but also introduce subjective scoring of a client manager, and the mode is more reasonable, but the subjectivity defect of the scoring of the client manager still cannot be avoided, so that how to realize objective prediction of financial data by using a relatively objective, accurate and efficient financial data prediction method is very important.
Aiming at the problem that in the prior art, financial data cannot be accurately predicted by relying on subjective judgment of a user or relying on objective judgment of personal user information, the implementation provides a financial data prediction method.
The method may be performed by an electronic device, a component in an electronic device, an integrated circuit, or a chip. The electronic device may be a mobile electronic device or a non-mobile electronic device. The mobile electronic device may be a mobile phone, a tablet computer, a notebook computer, a palm computer, a vehicle-mounted electronic device, a wearable device, an ultra mobile personal computer, or the like, and the non-mobile electronic device may be a server, a network attached memory, a personal computer, or the like, which is not particularly limited.
The method for predicting financial data according to the present invention is applicable to the scenario of predicting the amount of money paid by a customer, predicting the consumption level of the customer, etc., and is not limited in detail in this embodiment.
As shown in fig. 1, one of the flow charts of the financial data prediction method provided in the present embodiment is shown, and the method includes the following steps:
the target user is a customer who needs to make financial data prediction, and may be an individual customer.
The dialogue speech may be speech data generated by interaction between the target user and the client manager in the financial dialogue scene, or speech data generated by interaction between the target user and the electronic device in the financial dialogue scene, and the generation mode of the dialogue speech is not specifically limited in this embodiment. The dialogue voice can objectively reflect the consumption desire, consumption preference, excitation degree and the like of the target user.
The user information is used to characterize personal information of the user, including but not limited to revenue, age, home context, educational context, work context, offence records, historical consumption limits, historical purchasing behavior, etc., which is not specifically limited in this embodiment.
It should be noted that, the user information may be input by the target user in real time, or may be stored in advance by the target user.
Optionally, when the financial data prediction is required for the target user, the dialogue voice of the target user and the user information of the target user can be collected in real time; after the dialogue voice and the personal information of the target user are obtained, the dialogue voice and the personal information of the target user can be preprocessed, including but not limited to data cleaning, data normalization processing and the like, so that the quality of the data information is improved, and the follow-up more convenient and accurate financial data prediction is facilitated.
optionally, in order to accurately predict the financial data of the target user, the consumption preference of the target user may be classified according to the dialogue speech, and the consumption preference level of the target user may be objectively mined.
Here, the classification of the consumption preference of the target user may be implemented by a pre-trained classification model, or may be implemented by a pre-configured consumption preference class classification rule, which is not specifically limited in this embodiment.
Optionally, after the consumption preference level of the target user is obtained, financial data prediction can be performed on the basis of the consumption preference level, so that a payment amount prediction result and/or a consumption level prediction result of the target user are obtained, and the purposes that the dissolubility of the model can be effectively improved, the accuracy of financial data prediction is improved, a service person is further assisted in determining how much payment should be paid to the target user, accurate payment of the target user is achieved, and user experience is improved are achieved by adding a variable, namely the consumption preference level, on the basis of traditional modeling.
Here, the prediction of the financial data for the target user may be implemented by a pre-trained financial data prediction model, or may be implemented by a pre-configured financial data prediction rule, which is not specifically limited in this embodiment.
In addition, after the prediction result of the financial data is obtained, the prediction result can be confirmed manually, so that the situation that the prediction of the financial data is inaccurate due to abnormal values of the output result is avoided, namely, deviation of the prediction result is avoided.
According to the financial data prediction method, the consumption preference of the target user is automatically and objectively classified according to the dialogue voice, so that the influence of manual experience is avoided, and the accuracy of the consumption preference classification of the target user is effectively improved; and on the basis of user information, the consumption preference level of the target user is combined, so that the financial data prediction result of the target user is automatically and objectively obtained through various factors related to the financial data prediction result, and the prediction efficiency of the financial data is effectively improved while the comprehensiveness, objectivity and accuracy of the financial data prediction of the target user are improved.
In some embodiments, the step of classifying the consumption preference of the target user according to the dialogue speech in step 102 to obtain a consumption preference level of the target user further includes:
inputting the dialogue voice into a voice conversion model, and converting the dialogue voice from voice data into text data to obtain text information corresponding to the dialogue voice;
and inputting the text information into a classification model, and classifying the consumption preference of the target user to obtain the consumption preference grade of the target user.
The voice conversion model is used for converting voice data into text data; the input of the voice conversion model is dialogue voice, and the output is text information corresponding to the dialogue voice, so that the text information corresponding to the dialogue voice is used as the input of the next classification model, and the credit line prediction result and/or the consumption level prediction result of the target user are accurately predicted.
The speech conversion model may be specifically implemented using an open source deep learning framework, for example, by implementing speech to text through a flyweight deep learning framework (hereinafter also referred to as a PaddlePaddle framework), and may be specifically implemented by a paddlespech in a PaddlePaddle library, so as to improve the accuracy of speech to text.
FIG. 2 is a schematic diagram of a speech conversion model, including a basic platform, wherein the basic platform includes a PaddlePaddle framework, a speech recognition tool Kaldi, a computing tool Sclite, an audio processing library Sox, and a matrix computing library Openblas; the voice conversion model also comprises a common module, specifically comprises a paddle audio module PaddleAudio, an integrated package Utils and the like; also included in the speech conversion model is the PaddleSpech model and data set. The speech conversion model is pre-installed in a Python library in the electronic device.
Optionally, when the dialogue voice needs to be converted from voice data to text data, the API (Application Programming Interface ) of the PaddlePaddle is directly called, so that the PaddleSpech in the PaddlePaddle library can be used for converting the dialogue voice from voice data to text data, and text information corresponding to the dialogue voice is obtained.
Wherein the classification model is used to classify the consumption preferences of the target user into a plurality of classes. The classification model may be constructed based on a pre-trained model BERT (Bidirectional Encoder Representation from Transformers, pre-trained language characterization model) or other classification model.
The BERT model is a pre-training model based on a transducer, and is commonly used for tasks in the field of natural language processing such as text classification, wherein the model introduces a mask (also called mask) mechanism to mask languages and adopts a multi-head attention mechanism to improve the parallel training efficiency and effect.
As shown in fig. 3, a schematic structural diagram of the BERT model is shown; wherein the BERT model specifically includes an encoding layer (also referred to as an Encoder) and a decoding layer (also referred to as a Decoder); the Encoder consists of a multi-head attention, a standardized network and a feed-forward network, and the Decode consists of a classical transducer module. Input encoding is the Input of the encoder and Output encoding is the Input of the decoder. The input of the encoder at the bottom layer is word vector, and word embedding vector is obtained through word2vec algorithm.
Optionally, when the classification model is generated based on the BERT model construction, if text information corresponding to the dialogue speech is acquired, transfer learning can be achieved through the BERT model, that is, based on the text information corresponding to the dialogue speech, deep learning is performed through the BERT model to automatically extract text features, word vector mapping is established, encoding and decoding of the text features are achieved through a transducer module based on a self-attention mechanism, finally probability distribution that the consumption preference of the target user belongs to each preset consumption preference level is obtained through a normalized exponential function (also called a Softmax function), and therefore the preset consumption preference level corresponding to the highest probability is determined as the consumption preference level of the target user.
It should be noted that, the consumption preference level output by the classification model may be encoded according to 1 to 5 levels; illustratively, at a rating of 1, a rating of consumption preference is lowest (i.e., a rating of consumption preference is too aggressive) and at a rating of 5, a rating of consumption preference is highest (i.e., a rating of consumption preference is too aggressive) is too conservative.
In the prior art, depending on the process of communicating with a personal client by a client manager, the client manager manually marks the consumption preference of the client according to the communication content, which can cause the judgment result of the client manager to be too subjective and influence the accuracy of financial data prediction.
In contrast, in the embodiment, by introducing an automatic flow engine, all links of data preprocessing, feature engineering, modeling, model evaluation, model output and the like are automatically connected in series, so that the application threshold of a machine learning technology in banking industry is reduced, namely, consumption preference levels are intelligently obtained on the basis of a voice conversion model and a classification model, and the method is a process for actually helping objectification of a manual judgment result; although the initial stage of manually marking may require some labor, in the long term, the accuracy of the overall financial data prediction can be improved while the objectivity of the overall financial data prediction is improved.
Therefore, on one hand, in the embodiment, the consumption preference level of the user is predicted by constructing the voice conversion model and the classification model, compared with the traditional mode of performing consumption preference judgment on individual clients only by means of the manual experience of a client manager, the method is more objective, provides important basis and powerful support for financial data prediction, objectively improves the financial data prediction precision, and can achieve the purposes of cost reduction and efficiency enhancement. On the other hand, on the basis of modeling based on user information, the consumption preference level of the target user is determined, so that the problem that dialogue voice information between a client manager and an individual client is ignored in a traditional modeling mode is effectively solved, and the prediction accuracy of financial data is further improved.
In some embodiments, the classification model is trained based on the following steps:
acquiring text information corresponding to dialogue voice of a sample user and a consumption preference grade label of the sample user;
constructing a training data set according to the text information corresponding to the sample user and the consumption preference grade label of the sample user;
based on the training data set, fine tuning is carried out on parameters of a pre-training model to obtain the classification model; the pre-training model is generated based on BERT model construction.
The consumption preference level is used for describing the consumption preference of the user, and the level division can be specifically set according to actual requirements. Illustratively, as shown in Table 1, consumption preference levels can be specifically categorized into five levels of over-aggressive, more aggressive, ordinary, more conservative, and over-conservative.
TABLE 1 results of consumer preference ranking
The BERT model is a pre-training model which is pre-trained based on a public data set, and when the BERT model is applied to a downstream consumption preference grade classification task, the BERT model is required to be finely adjusted based on text information corresponding to a sample user and a consumption preference grade label of the sample user on the basis of pre-training, so that a classification model capable of accurately outputting the consumption preference grade of the user can be obtained, and further accurate classification of the consumption preference grade of the user is realized.
Optionally, the training step of the classification model specifically includes:
firstly, collecting dialogue voice of a sample user, and converting the dialogue voice of the sample user into corresponding text information based on a voice conversion model;
and then, marking text information corresponding to the sample user based on a plurality of preset consumption preference levels so as to acquire a consumption preference level label of the sample user.
It should be noted that, the obtaining manner of the consumption preference level label may be obtained by performing marking based on one or more marking objects and/or one or more marking rules, which is not specifically limited in this embodiment. When marking is performed based on a marking object or a marking rule, the marking result can be directly used as a final consumption preference grade label; when marking is performed based on multiple marking objects and/or multiple marking rules, the multiple consumption preference level labels in the marking result need to be fused and then used as final consumption preference level labels.
Then, taking text information corresponding to the sample user as a sample and taking a consumption preference grade label of the sample user as a sample label to construct a training data set;
And then, carrying out fine adjustment on parameter iteration of the BERT model based on the training data set until the BERT model meets the model training termination condition, such as the maximum iteration times or the classification precision meets the precision threshold value.
And then, constructing an acquired classification model according to the optimal BERT model obtained by the last training.
In this embodiment, through fine tuning training on the pre-training model, on one hand, the classification model obtained by training can identify the consumption preference level of the user more accurately, and on the other hand, the subjective problem of scoring by the client manager in the prior art can be overcome, objective consumption preference level classification is realized by using the classification model based on the BERT model, so that objective scoring realized based on training of a large amount of data is realized, and compared with the traditional client manager scoring method, the objective scoring is more objective, efficient and intelligent, and further the prediction result of financial data is more accurate.
In some embodiments, obtaining the consumer preference level tag of the sample user comprises:
acquiring a plurality of initial consumption preference grade labels of the sample user; the plurality of initial consumption preference grade labels are generated by marking the consumption preference grade of the sample user through a plurality of marking objects and/or a plurality of marking rules;
And fusing the multiple initial consumption preference grade labels of the sample user to obtain the consumption preference grade label of the sample user.
Optionally, the marking manner of the consumption preference level label of the sample user may be that the consumption preference level of the sample user is manually marked based on a plurality of marking objects to obtain a plurality of initial consumption preference level labels, and then the consumption preference level labels of the sample user are obtained by fusing the plurality of initial consumption preference level labels, and in an exemplary manner, the consumption preference level of the sample user is scored based on the manager of at least three risk management posts, and the scoring results of the manager of at least three risk management posts are fused to obtain the consumption preference level label of the sample user.
Or, manually marking the consumption preference level of the sample user according to various marking rules based on the same marking object to obtain various initial consumption preference level labels, and then fusing the various initial consumption preference level labels to obtain the consumption preference level labels of the sample user;
or, based on different marking objects, manually marking the consumption preference level of the sample user according to various marking rules to obtain various initial consumption preference level labels, and then fusing the plurality of initial consumption preference level labels to obtain the consumption preference level label of the sample user. The marking manner of the consumption preference level label of the sample user is not particularly limited in this embodiment.
The fusion manner may be to average a plurality of initial consumption preference level labels or input a plurality of initial consumption preference level labels into a fusion model, which is not specifically limited in this embodiment.
In this embodiment, by fusing multiple initial consumption preference level labels of the sample user, the obtained consumption preference level labels of the sample user are more objective, so as to train and obtain a classification model capable of dividing the consumption preference level of the user more objectively and accurately, and further improve the accuracy of the financial data prediction result of the user.
In some embodiments, the constructing a training data set according to the text information corresponding to the sample user and the consumption preference level label of the sample user includes:
carrying out data enhancement on the text information corresponding to the sample user;
constructing the training data set according to the enhanced text information and the consumption preference grade label of the sample user;
wherein the data enhancement includes modifying the target vocabulary.
Data enhancement is often used for increasing the data volume so as to improve the diversity of the data and the robustness of a model, and is widely applied to the fields of computer vision, natural language processing and the like.
The target vocabulary may be non-key vocabulary in the text information corresponding to the sample user, i.e. vocabulary irrelevant to the financial prediction result.
Optionally, in the process of constructing the training data set, text information corresponding to the sample user can be multiplied by a data enhancement technology, so that the training data set acquired by the data enhancement technology contains richer sample data, thereby greatly improving the sample data volume, preparing for subsequent classification model training and fine tuning, and further improving the classification model training effect; and the consumption preference grade labels corresponding to the text information before and after data enhancement are consistent, so that the time for manual marking can be greatly saved, the labor cost is saved, and the method is a typical application scene of the artificial intelligence technology in the financial field for reducing the end-to-end landing cost.
In some embodiments, after classifying the consumption preference of the target user according to the dialogue speech, the method further includes:
and expanding the training data set based on the text information corresponding to the target user and the consumption preference level of the target user.
Optionally, after obtaining the consumption preference level of the target user, the training data set may be further expanded in real time based on the text information corresponding to the target user and the consumption preference level of the target user, so as to further fine-tune the classification model based on the expanded training data set, so as to further improve the classification precision of the classification model, and further make the prediction result of the financial data of the target user more accurate.
It should be noted that the fine tuning frequency may be set according to actual requirements, for example, with a fine tuning frequency of one day or week, and the model effect is improved by further fine tuning the classification model based on the extended training data set acquired daily or weekly.
In some embodiments, the step of predicting the financial data of the target user according to the consumption preference level and the user information in step 103 further comprises:
inputting the consumption preference level and the user information into a financial data prediction model, and carrying out attribution analysis and financial data prediction on the consumption preference level and the user information to obtain a financial data prediction result of the target user;
the financial data prediction model may be generated based on a regression model and/or tree model construction.
Optionally, in order to more deeply mine the influence of the consumption preference level and the user information on the financial data prediction result, so as to more conveniently and accurately obtain the financial data prediction result, a financial data prediction model may be preconfigured to perform attribution analysis and financial data prediction on the consumption preference level and the user information based on the financial data prediction model, so as to obtain the financial data prediction result of the target user.
The attribution analysis may be performed based on the financial data prediction model for the consumption preference level factor and each user information factor in the user information, such as income, age, family background, education background, work background, default record, historical consumption amount, historical purchasing behavior, and the like, so as to objectively determine contribution degree of each factor to the financial data prediction result output by the financial data prediction model. Among other things, the tree model can be attributed based on Xgboost (eXtreme Gradient Boosting, extreme gradient lifting) and shape (Shapley Additive exPlanation, interpretable machine learning).
It should be noted that, attribution molecules are used for assisting in judging importance of consumption preference levels in different periods and ordering among factors, so that prediction of financial data by a financial data prediction model is better assisted, judgment and verification of financial data prediction results are manually performed, and accuracy of the financial data prediction results is further improved.
The step of predicting the financial data may be to determine coefficients of each factor according to the contribution degree of each factor to the predicted result of the financial data, so as to fusion predict each factor to obtain the predicted result of the financial data of the target user. The financial data prediction may be implemented based on a regression model, which may be a linear regression model or a nonlinear regression model, which is not specifically limited in this embodiment.
In this embodiment, the contribution degree of each factor to the predicted result of the financial data is more accurately mined by performing attribution analysis on the consumption preference level and the user information based on the predicted model of the financial data, and fusion prediction is performed on each factor according to the contribution degree of each factor to the predicted result of the financial data, so as to obtain the predicted result of the financial data of the target user more accurately.
The following describes the complete flow of the financial data prediction method provided in this embodiment with a specific example in conjunction with fig. 4 and 5.
As shown in fig. 4 and fig. 5, two and three flow diagrams of the financial data prediction method are respectively shown, wherein the structured data is user information of the user, and the unstructured data is dialogue voice of the user. The complete flow of the financial data prediction method specifically comprises the following steps:
Step a), collecting dialogue voice of a target user and dialogue voice of a sample user; the method for collecting dialogue voice can be recording when a customer manager and a user are in dialogue under a financial dialogue scene, and the user needs to be prompted that communication is to be recorded before recording, so that the dialogue voice is used as an input source of unstructured data;
step b), after recording collection is completed, uploading the recording file to a corresponding file directory in a database;
step c), the electronic equipment reads the dialogue voice of the target user and the dialogue voice of the sample user in the corresponding file catalogue in the database, invokes the PaddleSpech under the PaddlePaddle framework to perform voice data to text data operation on the dialogue voice to obtain text information corresponding to the dialogue voice, and stores the text information into the target file, such as a temp_result folder;
marking text information corresponding to dialogue voice of the sample user, and obtaining consumption preference grade labels of the sample user so as to provide a training data set for the classification model; it should be noted that, in the process of constructing the training data set, marking can be performed based on at least three operators in risk management post, and then an average value of marking results of at least three operators is taken as a final consumption preference grade label of the sample user; in addition, the text information corresponding to the sample user can be subjected to data enhancement by adopting a data enhancement technology, so that the training data set contains more and richer training samples;
Step e), fine tuning the BERT model based on the training data set to obtain a classification model, classifying text information corresponding to the target user in the temp_result folder based on the classification model, and outputting classification results to an output text folder, such as an output_prob folder, to obtain the consumption preference level of the target user;
step f), modeling by adopting a financial data prediction model (constructed and generated based on a regression model and a tree model) based on consumption preference levels of the target users and user information (such as income, age, family background, educational background, work background, default records, historical consumption amount, historical purchasing behavior and the like), so as to predict and obtain a payment amount prediction result and/or consumption level prediction result of the target users;
and g), after obtaining the payoff amount prediction result and/or the consumption level prediction result output by the model, manually comparing and rechecking the payoff amount prediction result and/or the consumption level prediction result so as to avoid deviation of model prediction.
In summary, the financial data prediction method provided in the embodiment has the following advantages:
one of the advantages is that by introducing an automatic machine learning engine to convert voice into text, classifying the text, modeling machine learning, attribution analysis, data enhancement technology and outputting model results to the front of manual judgment, each process can be connected in series by introducing the automatic machine learning technology, so that automatic training and automatic model tuning are realized, the labor cost is reduced, and the prediction precision is improved;
And secondly, introducing a classification model to automatically identify dialogue voices of users to obtain consumption preference levels, and modeling by combining the consumption preference levels on the basis of user information to further enable prediction to obtain financial data more accurately.
And thirdly, the dialogue voice between the client manager and the individual client is also an important asset in practice, the consumption preference level predicted by the classification model is widely used in the individual financial departments of banking industry, not only can be used for predicting the payoff amount and consumption level of the individual client, but also can be used for compliance checking, providing fields of later training and the like, and providing important data information for asset precipitation of financial data.
The financial data prediction apparatus provided by the present invention will be described below, and the financial data prediction apparatus described below and the financial data prediction method described above may be referred to correspondingly to each other.
As shown in fig. 6, the apparatus for predicting financial data provided in this embodiment includes:
the acquiring module 601 is configured to acquire dialogue speech of a target user and user information of the target user; the dialogue speech is generated by the target user in a financial dialogue scene;
The classification module 602 is configured to classify consumption preferences of the target user according to the dialogue speech, so as to obtain consumption preference levels of the target user;
the prediction module 603 is configured to obtain a financial data prediction result of the target user according to the consumption preference level and the user information; the financial data prediction results comprise a payoff amount prediction result and/or a consumption level prediction result.
According to the financial data prediction device provided by the embodiment, the consumption preference of the target user is automatically and objectively classified according to the dialogue voice, so that the influence of manual experience is avoided, and the accuracy of the consumption preference classification of the target user is effectively improved; and on the basis of user information, the consumption preference level of the target user is combined, so that the financial data prediction result of the target user is automatically and objectively obtained through various factors related to the financial data prediction result, and the prediction efficiency of the financial data is effectively improved while the comprehensiveness, objectivity and accuracy of the financial data prediction of the target user are improved.
In some embodiments, classification module 602 is specifically configured to:
inputting the dialogue voice into a voice conversion model, and converting the dialogue voice from voice data into text data to obtain text information corresponding to the dialogue voice;
And inputting the text information into a classification model, and classifying the consumption preference of the target user to obtain the consumption preference grade of the target user.
In some embodiments, the apparatus further comprises a training module, in particular for:
acquiring text information corresponding to dialogue voice of a sample user and a consumption preference grade label of the sample user;
constructing a training data set according to the text information corresponding to the sample user and the consumption preference grade label of the sample user;
based on the training data set, fine tuning is carried out on parameters of a pre-training model to obtain the classification model; the pre-training model is generated based on BERT model construction.
In some embodiments, the training module is further to:
acquiring a plurality of initial consumption preference grade labels of the sample user; the plurality of initial consumption preference grade labels are generated by marking the consumption preference grade of the sample user through a plurality of marking objects and/or a plurality of marking rules;
and fusing the multiple initial consumption preference grade labels of the sample user to obtain the consumption preference grade label of the sample user.
In some embodiments, the training module is further to:
Carrying out data enhancement on the text information corresponding to the sample user;
constructing the training data set according to the enhanced text information and the consumption preference grade label of the sample user;
wherein the data enhancement includes modifying the target vocabulary.
In some embodiments, the apparatus further comprises a data expansion module, specifically configured to:
and expanding the training data set based on the text information corresponding to the target user and the consumption preference level of the target user.
In some embodiments, the prediction module 603 is specifically configured to:
inputting the consumption preference level and the user information into a financial data prediction model, and carrying out attribution analysis and financial data prediction on the consumption preference level and the user information to obtain a financial data prediction result of the target user;
the financial data prediction model may be generated based on a regression model and a tree model construction.
Fig. 7 illustrates a physical schematic diagram of an electronic device, as shown in fig. 7, which may include: a processor (processor) 701, a communication interface (Communications Interface) 702, a memory (memory) 703 and a communication bus 704, wherein the processor 701, the communication interface 702 and the memory 703 communicate with each other through the communication bus 704. The processor 701 may invoke logic instructions in the memory 703 to perform a financial data prediction method comprising: acquiring dialogue voice of a target user and user information of the target user; the dialogue speech is generated by the target user in a financial dialogue scene; classifying the consumption preference of the target user according to the dialogue voice to obtain the consumption preference level of the target user; acquiring a financial data prediction result of the target user according to the consumption preference level and the user information; the financial data prediction results comprise a payoff amount prediction result and/or a consumption level prediction result.
Further, the logic instructions in the memory 703 may be implemented in the form of software functional units and may be stored in a computer readable storage medium when sold or used as a stand alone product. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
In another aspect, the present invention also provides a computer program product comprising a computer program storable on a non-transitory computer readable storage medium, the computer program when executed by a processor being capable of performing the financial data prediction method provided by the methods above, the method comprising: acquiring dialogue voice of a target user and user information of the target user; the dialogue speech is generated by the target user in a financial dialogue scene; classifying the consumption preference of the target user according to the dialogue voice to obtain the consumption preference level of the target user; acquiring a financial data prediction result of the target user according to the consumption preference level and the user information; the financial data prediction results comprise a payoff amount prediction result and/or a consumption level prediction result.
In yet another aspect, the present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, is implemented to perform the method of predicting financial data provided by the above methods, the method comprising: acquiring dialogue voice of a target user and user information of the target user; the dialogue speech is generated by the target user in a financial dialogue scene; classifying the consumption preference of the target user according to the dialogue voice to obtain the consumption preference level of the target user; acquiring a financial data prediction result of the target user according to the consumption preference level and the user information; the financial data prediction results comprise a payoff amount prediction result and/or a consumption level prediction result.
The apparatus embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.
From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course may be implemented by means of hardware. Based on this understanding, the foregoing technical solution may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method described in the respective embodiments or some parts of the embodiments.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (10)
1. A method of predicting financial data, comprising:
acquiring dialogue voice of a target user and user information of the target user; the dialogue speech is generated by the target user in a financial dialogue scene;
classifying the consumption preference of the target user according to the dialogue voice to obtain the consumption preference level of the target user;
acquiring a financial data prediction result of the target user according to the consumption preference level and the user information; the financial data prediction results comprise a payoff amount prediction result and/or a consumption level prediction result.
2. The financial data prediction method according to claim 1, wherein classifying the consumption preference of the target user according to the dialogue speech to obtain the consumption preference level of the target user comprises:
inputting the dialogue voice into a voice conversion model, and converting the dialogue voice from voice data into text data to obtain text information corresponding to the dialogue voice;
and inputting the text information into a classification model, and classifying the consumption preference of the target user to obtain the consumption preference grade of the target user.
3. The method of claim 2, wherein the classification model is trained based on the steps of:
acquiring text information corresponding to dialogue voice of a sample user and a consumption preference grade label of the sample user;
constructing a training data set according to the text information corresponding to the sample user and the consumption preference grade label of the sample user;
based on the training data set, fine tuning is carried out on parameters of a pre-training model to obtain the classification model; the pre-training model is generated based on BERT model construction.
4. The financial data prediction method of claim 3, wherein obtaining a consumer preference level label for the sample user comprises:
acquiring a plurality of initial consumption preference grade labels of the sample user; the plurality of initial consumption preference grade labels are generated by marking the consumption preference grade of the sample user through a plurality of marking objects and/or a plurality of marking rules;
and fusing the multiple initial consumption preference grade labels of the sample user to obtain the consumption preference grade label of the sample user.
5. The financial data prediction method according to claim 3, wherein the constructing a training data set according to the text information corresponding to the sample user and the consumption preference level label of the sample user includes:
Carrying out data enhancement on the text information corresponding to the sample user;
constructing the training data set according to the enhanced text information and the consumption preference grade label of the sample user;
wherein the data enhancement includes modifying the target vocabulary.
6. A financial data prediction method according to claim 3, wherein after classifying the consumption preference of the target user according to the dialogue speech to obtain the consumption preference level of the target user, the method further comprises:
and expanding the training data set based on the text information corresponding to the target user and the consumption preference level of the target user.
7. The financial data prediction method according to any one of claims 1 to 6, wherein predicting the financial data of the target user based on the consumption preference level and the user information comprises:
inputting the consumption preference level and the user information into a financial data prediction model, and carrying out attribution analysis and financial data prediction on the consumption preference level and the user information to obtain a financial data prediction result of the target user;
The financial data prediction model may be generated based on a regression model and a tree model construction.
8. A financial data prediction apparatus, comprising:
the system comprises an acquisition module, a processing module and a processing module, wherein the acquisition module is used for acquiring dialogue voice of a target user and user information of the target user; the dialogue speech is generated by the target user in a financial dialogue scene;
the classification module is used for classifying the consumption preference of the target user according to the dialogue voice to obtain the consumption preference grade of the target user;
the prediction module is used for acquiring a financial data prediction result of the target user according to the consumption preference level and the user information; the financial data prediction results comprise a payoff amount prediction result and/or a consumption level prediction result.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the financial data prediction method of any one of claims 1 to 7 when the program is executed by the processor.
10. A non-transitory computer readable storage medium having stored thereon a computer program, which when executed by a processor implements the financial data prediction method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310093442.4A CN116308735A (en) | 2023-02-06 | 2023-02-06 | Financial data prediction method, device, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310093442.4A CN116308735A (en) | 2023-02-06 | 2023-02-06 | Financial data prediction method, device, electronic equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116308735A true CN116308735A (en) | 2023-06-23 |
Family
ID=86795104
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310093442.4A Pending CN116308735A (en) | 2023-02-06 | 2023-02-06 | Financial data prediction method, device, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116308735A (en) |
-
2023
- 2023-02-06 CN CN202310093442.4A patent/CN116308735A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110377911B (en) | Method and device for identifying intention under dialog framework | |
| US20150142446A1 (en) | Credit Risk Decision Management System And Method Using Voice Analytics | |
| CN113468296B (en) | Model self-iteration type intelligent customer service quality inspection system and method capable of configuring business logic | |
| CN106709804A (en) | Interactive wealth planning consulting robot system | |
| CN108549697A (en) | Information-pushing method, device, equipment based on semantic association and storage medium | |
| CN107480149A (en) | Answer in question answering system provides method and device | |
| CN112732871A (en) | Multi-label classification method for acquiring client intention label by robot | |
| CN114722839A (en) | Man-machine collaborative dialogue interaction system and method | |
| GB2521050A (en) | Credit risk decision management system and method using voice analytics | |
| CN110399473A (en) | The method and apparatus for determining answer for customer problem | |
| CN107766560B (en) | Method and system for evaluating customer service flow | |
| CN112132679A (en) | Product recommendation method and device | |
| CN117435471A (en) | Method, device, equipment, storage medium and program product for recommending test cases | |
| CN116842263A (en) | Training processing method and device for intelligent question-answering financial advisor model | |
| CN116738198A (en) | Information identification method, device, equipment, medium and product | |
| CN116757835A (en) | Method and device for monitoring transaction risk in credit card customer credit | |
| CN116883888A (en) | Bank counter service problem tracing system and method based on multi-mode feature fusion | |
| CN117034955A (en) | Telephone traffic text intention recognition method and device | |
| CN116308735A (en) | Financial data prediction method, device, electronic equipment and storage medium | |
| KR20230116143A (en) | Counseling Type Classification System | |
| CN110110048B (en) | Query guiding method and device | |
| KR102507810B1 (en) | Voice-based sales information extraction and lead recommendation method using artificial intelligence, and data analysis apparatus therefor | |
| CN111932018B (en) | Bank business performance contribution information prediction method and device | |
| CN117009454A (en) | Data response processing method, system, equipment and storage medium | |
| CN117874225A (en) | Customer complaint work order intention recognition method and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |