[ summary of the invention ]
In view of this, the embodiment of the present invention provides a financial wind control method based on a block chain.
The embodiment of the invention provides a financial wind control method based on a block chain, which is applied to a block chain network, wherein the block chain network comprises a monitoring node, a prediction node, a consensus node, a user node and an expert node, and the method comprises the following steps:
s1, when a consensus node receives task request information broadcasted by a user node in a block chain, comparing the task request information with a monitoring list in the current period, wherein the task request information comprises a financial monitoring target, an intelligent reward payment contract for agreeing a reward rule and an issued encrypted digital currency amount;
s2, if the financial monitoring target is repeated with the monitoring list, the consensus node sends refusal information to the user node, and if not, the consensus node sends a query request to the monitoring node;
s3, the monitoring node acquires comment data of the fusion monitoring target in a preset time period from a preset financial data source platform, preprocesses the comment data, calculates an emotion index E, and if the emotion index E is larger than an emotion index threshold E 0 If yes, a prediction request is sent to the prediction node;
s4, the forecasting node forecasts the price change trend of the financial monitoring target based on the constructed investment forecasting model, and broadcasts the forecasting result with a timestamp in the block network;
and S5, the user node acquires the prediction result and issues a corresponding amount of encrypted digital currency to the prediction node based on the intelligent reward payment contract.
The above-described aspect and any possible implementation manner further provide an implementation manner, where before step S1, the method further includes:
s10, if the update period is reached, the consensus node receives an investment list which is encrypted and shared by the user node in the blockchain network;
s20, the consensus node enables the investment index I to be larger than the investment index threshold I 0 The investment target is screened to be used as a financial monitoring target, and a monitoring list is generated and sent to a monitoring node;
s30, the monitoring node acquires the comment data of the fusion monitoring target in a preset time period from the preset financial data source platform, preprocesses the comment data, calculates an emotion index E, and if the emotion index E is larger than an emotion index threshold E 0 If yes, a prediction request is sent to the prediction node;
and S40, the forecasting node forecasts the price change trend of the financial monitoring target based on the constructed investment forecasting model, and broadcasts the forecasting result with a timestamp in the block network.
As to the above-mentioned aspect and any possible implementation manner, there is further provided an implementation manner, where the step S10 specifically includes:
s101, the consensus node and the user node respectively send a public key corresponding to the private key to a block chain;
s102, calculating a hash value corresponding to investment data by a user node to serve as a private key, encrypting the investment data through the private key, performing secondary encryption through a public key of a consensus node, and sending encrypted content to a block chain, wherein the investment data comprise an investor ID, an investment target name and a target amount;
s103, the consensus node verifies the investment data after the public key is encrypted according to the private key of the consensus node, and then decrypts the encrypted content through the public key of the user node to obtain the investment data;
s104, the consensus node calculates a hash value corresponding to the investor ID and the name of the investment target, encrypts the investment data through a private key, and then sends the encrypted content to a block chain.
The above-mentioned aspect and any possible implementation manner further provide an implementation manner, and the investment index I in the step S20 is calculated according to the following formula:
wherein I (x) represents an investment function, x represents the serial number of an investment target, alpha x Amount, alpha, representing the xth target 0 Total amount of investment target, m x Representing the number of the xth investment target, m 0 Representing the total number of investment targets; w is a 1 ,w 2 Represents a weight satisfying w 1 ,w 2 ∈[0,1]And w 1 +w 2 =1。
The above-mentioned aspect and any possible implementation manner further provide an implementation manner, in the step S30, the review data is preprocessed, which includes a first priority processing and a second priority processing, where the first priority processing is as follows:
s301, deleting messy codes, blanks and picture comments;
s302, when the same IP user is in T 1 The same comment number M is more than or equal to the same comment threshold value M when being published on the same platform in time 1 When the comment is deleted, redundant comments are deleted, and only one comment is reserved;
s303, when the same IP user is in T 2 More than comment number M is more than or equal to comment number threshold value M when being published on the same platform in time 2 When the same IP user is at T 3 The number of more comments M and more than the number of comments M are published on different platforms within time is more than or equal to the threshold value M of the number of comments 3 When it is time, delete all its comments, where T 3 ≥T 2 ≥T 1 And M is 2 ≥M 3 。
The above-described aspect and any possible implementation further provide an implementation, where the second priority processing includes:
s304, obtaining the comment text, performing word segmentation and stop word and preposition word removing processing to obtain keywords;
s305, screening the keywords according to the screening model F (x), wherein the deletion value is smaller than the threshold value F 0 Wherein the screening model F (x) is defined as follows:
wherein F (x) represents a filtering function, x represents a serial number of a keyword, q x Representing the number of times of occurrence of the xth keyword in the comment, c representing the total number of keywords in the comment containing the xth keyword, N representing the total number of comments, and N representing the number of comments of the keyword;
s306, secondary classification is carried out on the screened keywords through a pre-trained good classification model, and positive comments or negative comments are obtained.
The above-mentioned aspect and any possible implementation manner further provide an implementation manner, and the emotion index E in step S30 is calculated as follows:
wherein E represents the sentiment index, g 1 The growth rate, p, of positive comments representing the current update period 1 Number of positive comments,/ 1 Indicates positive comment, i 2 Indicates positive comment on the number of persons stepped on, g 2 Growth rate of negative comments, p, representing the current update period 2 Number of negative comments,/ 3 Number of negative comments,/ 4 Indicates the number of passive comment steps, and t indicates the time in hours.
The above-described aspects and any possible implementations further provide an implementation, and the method further includes:
the monitoring node calculates a divergence index D, if the emotion index E is larger than an emotion index threshold E 0 And divergence index D > divergence index threshold D 0 Sending an assessment request to the expert node, wherein the divergence index D is calculated according to the following formula:
wherein D represents a divergence index, m * Is shown asNumber of task requests of user node in pre-update period, g 1 The growth rate, p, of positive comments representing the current update period 1 Number of positive comments,/ 1 Indicates positive comment, i 2 Indicates positive comment on the number of persons stepped on, g 2 Growth rate of negative comments, p, representing the current update period 2 Number of negative comments,/ 3 Indicates negative comment number,/ 4 Representing the number of people on which negative comments are stepped;
the expert node obtains the prediction result of the prediction node, and adds the expert evaluation suggestion broadcast in the block network.
The above-mentioned aspects and any possible implementation manner further provide an implementation manner, where the predicting node predicts the price change trend of the financial monitoring target based on the constructed investment prediction model in step S40, specifically including:
obtaining historical trading data of a plurality of stocks, modeling the stock trading process as a Markov decision process, and specifically comprising the following steps:
the state is represented by s, which is the environment state and the stock price information generated by the behavior strategy;
action a represents, which includes buy, hold, and sell;
r (s, a, s) for reward * ) Represents when taking action a at state s and arriving at new state s * The change of the time investment value, namely the single step reward value fed back by the environment, wherein the investment value is the total value of the stock value and the balance;
defining future returns R
t A weighted sum of the prize values earned for all actions from the current state to the future state,
wherein the content of the first and second substances,
T denotes the total amount of the sample, γ
i-t Represents the reward discount coefficient of the t sample to the i sample, r(s)
i ,a
i ,s
i+1 ) Indicates when in state s
i Taking action a
i And reaches a new state s
i+1 A change in the time investment price;
the strategy is expressed by pi(s), and is a stock trading strategy of a state s, namely the probability distribution of the action a in the state s and the action to be taken next;
defining a state-action value function Q π (s, a), which is the expected reward achieved by action a when policy π is followed in state s;
obtaining an optimal state-action value function Q through a Bellman equation π (s t ,a t ):
Wherein Q π (s t ,a t ) Is a specific state s t According to a specific strategy π Performing action a t And future rewards are expected by returning r(s) t ,a t ,s t+1 Is expected to add the next state s t+1 Calculated from expected returns of; e represents expectation;
simultaneous, state-action value function Q π (s t ,a t ) The update process can be represented as follows:
δ(t)=r(s t ,a t ,s t+1 )-Q π (s t ,a t ),
wherein the initial Q value before learning by environment is set to 0, a represents a learning rate for adjusting the variation range from one experiment to the next experiment, a + =1,a - =0, δ (t) represents the prediction error, being the expected return Q π (s t ,a t ) And the actual return r(s) t ,a t ,s t+1 ) The difference between them;
using greedy actions a t+1 To maximize the state s t+1 Q(s) of t+1 ,a t+1 ) The following were used:
the DNN is introduced into the framework of Q-learning, consisting of an Online network that uses a Q function Q (s, a, θ) with a weight θ to approximate an optimal state-action value function Q and a Target network π (s t ,a t ) (ii) a Target network usage with weight θ - Q function Q (s, a, theta) - ) To improve the performance of the whole network, after a certain number of rounds, the weight theta of the Online network is copied to update the weight theta of the Target network - Updating the weight theta of the Online network by using a gradient descent method to obtain a minimum loss function:
wherein L represents a loss function, r represents a reward value, theta and theta' represent network weights,
represents the target Q function value, Q (s, a, θ) represents the predicted Q function value, γ represents the discount factor;
and predicting the price change trend of the financial monitoring target by using the trained deep learning network.
The above-mentioned aspect and any possible implementation manner further provide an implementation manner, and the step S5 wins the payment intelligence contract, specifically including:
the user node acquires the prediction result and pays half amount of encrypted digital currency to the prediction node;
and when the current updating period is finished, judging whether the prediction is correct or not, and if the prediction is correct, paying the other half amount of encrypted digital currency to the prediction node.
One of the above technical solutions has the following beneficial effects:
the financial wind control method based on the block chain is provided in the method of the embodiment of the invention, so that the real-time monitoring of the financial network public sentiment is realized, the possible market fluctuation can be laid out and emergent in advance by investors, the possible fluctuating financial targets can be predicted by establishing a model, the prediction accuracy can be improved, the subjective and popular psychology of the investors is avoided, rationality and reference are provided for the investors, and the investment risk of the stocks, the base people and investment institutions is effectively reduced; the initiative of processing tasks, updating and perfecting the prediction model can be improved.
[ detailed description ] A
In order to make 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 specific embodiments and corresponding drawings. It is to be understood that the described embodiments are merely exemplary of the invention, and not restrictive of the full scope of the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Referring to fig. 1, fig. 1 is a flowchart illustrating S1-S5 of a block chain-based financial wind control method according to an embodiment of the present invention. As shown in fig. 1, the method is applied to a blockchain network including a monitoring node, a prediction node, a consensus node, a user node, and an expert node, and includes:
s1, when a consensus node receives task request information broadcasted by a user node in a block chain, comparing the task request information with a monitoring list of a current period, wherein the task request information comprises a financial monitoring target, an award payment intelligent contract for agreeing an award rule and an issued encrypted digital currency amount;
s2, if the financial monitoring target is repeated with the monitoring list, the consensus node sends refusal information to the user node, and otherwise, sends a query request to the monitoring node;
s3, the monitoring node acquires the comment data of the fusion monitoring target in a preset time period from the preset financial data source platform, pre-processes the comment data, calculates an emotion index E, and if the emotion index E is larger than an emotion index threshold E 0 If yes, a prediction request is sent to the prediction node;
s4, the forecasting node forecasts the price change trend of the financial monitoring target based on the constructed investment forecasting model, and broadcasts the forecasting result with a timestamp in the block network;
and S5, the user node acquires the prediction result and issues a corresponding amount of encrypted digital currency to the prediction node based on the intelligent reward payment contract.
The consensus node is used for receiving task request information of a user node, and the task requested by the user node is price change prediction of a financial monitoring target, such as the rise and fall of a certain stock or fund, so that reference is provided for investment of a user; the monitoring list of the current period refers to that the system monitors the financial targets in the list and provides information for the user nodes free of charge, so that if the financial monitoring targets requested by the user nodes are repeated with the monitoring list, the user node request is directly rejected without extra waste of resources, and extra expenditure of the user nodes is avoided; if the financial monitoring target and the monitoring list are not repeated, sending an inquiry request to a monitoring node, and acquiring comment data of the financial monitoring target in a preset time period by the monitoring node from a preset financial data source platform, wherein the financial data source platform can be, for example, a stock forum, an app containing fund business and the like, and the acquired comment data is used for monitoring financial network public opinions and enabling investors to make layout and emergency in advance for possible market fluctuation; if the emotion index E is larger than the emotion index threshold E0, a prediction request is sent to the prediction node, the goal with higher heat can be monitored, and the workload of the monitoring node and the prediction node can be reduced by setting the threshold; the forecasting node forecasts the price change trend of the financial monitoring target based on the established investment forecasting model, and forecasts the financial target which is likely to fluctuate by establishing the model, so that on one hand, the forecasting accuracy can be improved, and the forecasting model becomes a good financial management means for investors, on the other hand, the subjective crowd psychology of the investors can be avoided, rationality and reference are provided for the investors, and the investment risk of stockholders, basic people and investment institutions is effectively reduced; the user node obtains the prediction result, and sends the corresponding amount of encrypted digital currency to the prediction node based on the reward payment intelligent contract, through a reward mechanism, the enthusiasm of the prediction node for processing tasks, updating and perfecting a prediction model can be improved, and the safety and the enthusiasm of payment can be guaranteed through constraint of the intelligent contract.
In addition, step S5 wins the incentive payment intelligent contract, which specifically includes:
the user node acquires the prediction result and pays half amount of encrypted digital currency to the prediction node;
and when the current updating period is finished, judging whether the prediction is correct, and if so, paying the other half amount of encrypted digital currency to the prediction node.
By setting the reward rule through the intelligent contract, the efficient incentive to the prediction node can be realized. It should be noted that investment prediction models built by different prediction nodes may be shared, or may be independent after update training, or even may be independent. Screening mechanisms can be provided for the predicted nodes, for example, the predicted nodes can provide uplink registration, and the predicted nodes are rewarded differently or even eliminated according to the prediction accuracy of the predicted nodes.
Referring to fig. 2, fig. 2 is a flowchart illustrating S10-S40 of a financial wind control method based on a block chain according to an embodiment of the present invention. As shown in fig. 2, the method further includes, before step S1:
s10, if the update period is reached, the consensus node receives an investment list which is encrypted and shared by the user node in the blockchain network;
s20, the consensus node enables the investment index I to be larger than an investment index threshold value I 0 The investment target is screened to be used as a financial monitoring target, and a monitoring list is generated and sent to a monitoring node;
s30, the monitoring node acquires the comment data of the fusion monitoring target in a preset time period from the preset financial data source platform, preprocesses the comment data, calculates an emotion index E, and if the emotion index E is larger than an emotion index threshold E 0 If yes, a prediction request is sent to the prediction node;
and S40, the forecasting node forecasts the price change trend of the financial monitoring target based on the constructed investment forecasting model, and broadcasts the forecasting result with a timestamp in the block network.
Wherein, the calculation formula of the investment index I in the step S20 is as follows:
wherein I (x) represents an investment function, x represents the serial number of an investment target, alpha x Amount of money, alpha, representing the xth investment target 0 Total amount of investment target, m x Representing the number of the xth investment target, m 0 Representing the total number of investment targets; w is a 1 ,w 2 Represents a weight satisfying w 1 ,w 2 ∈[0,1]And w 1 +w 2 =1。
In addition, the emotion index E in step S30 is calculated as follows:
wherein E represents an emotion index, g 1 The growth rate, p, of positive comments representing the current update period 1 Number of positive comments,/ 1 Indicates positive comment number of praise,/ 2 Indicates the number of persons stepping on the floor according to positive comments, g 2 Growth rate of negative comments, p, representing the current update period 2 Number of negative comments,/ 3 Number of negative comments,/ 4 Representing the number of passive commentary steps and t representing the time in hours.
In addition, the predicting node predicts the price change trend of the financial monitoring target based on the constructed investment prediction model in the step S40, and specifically includes:
obtaining historical stock trading data, modeling a stock trading process as a Markov decision process, and specifically comprising the following steps:
the state is represented by s, which is the environment state and the stock price information generated by the behavior strategy;
action a represents, which includes buy, hold, and sell;
r (s, a, s) for bonus * ) Indicates that it is when taking action a at state s and reaching a new state s * The change of the time investment value, namely the single step reward value fed back by the environment, wherein the investment value is the total value of the stock value and the balance;
defining future returns R
t A weighted sum of the prize values earned for all actions from the current state to the future state,
wherein, the first and the second end of the pipe are connected with each other,
T denotes the total amount of the sample, γ
i-t Represents the reward discount coefficient of the t sample to i sample, r(s)
i ,a
i ,s
i+1 ) Indicates when in state s
i Taking action a
i And reaches a new state s
i+1 A change in the time investment price;
the strategy is expressed by pi(s), and is a stock trading strategy of a state s, namely the probability distribution of the action a in the state s, and the action to be taken next;
defining a state-action value function Q π (s, a), which is the expected reward achieved by action a when policy π is followed in state s;
obtaining an optimal state-action value function Q through a Bellman equation π (s t ,a t ):
Wherein Q is π (s t ,a t ) Is a specific state s t According to a specific strategy π Performing action a t And the future reward obtained is expected by the reward r(s) t ,a t ,s t+1 Is expected to add the next state s t+1 Calculated from expected returns of; e represents expectation;
simultaneous, state-action value function Q π (s t ,a t ) The update process can be represented as follows:
δ(t)=r(s t ,a t ,s t+1 )-Q π (s t ,a t ),
wherein the initial Q value before learning by the environment is set to 0, a represents a learning rate for adjusting the range of change from one trial to the next, a + =1,a - =0, δ (t) represents the prediction error, being the expected return Q π (s t ,a t ) And the actual return r(s) t ,a t ,s t+1 ) The difference therebetween;
using greedy action a t+1 To maximize the state s t+1 Q(s) of (2) t+1 ,a t+1 ) The following:
the DNN is introduced into the framework of Q-learning, consisting of an Online network that uses a Q function Q (s, a, θ) with a weight θ to approximate an optimal state-action value function Q and a Target network π (s t ,a t ) (ii) a Target network usage with weight θ - Q function of (s, a, theta) - ) To improve the performance of the whole network, after a certain number of rounds, the weight theta of the Online network is copied to update the weight theta of the Target network - Updating the weight theta of the Online network by using a gradient descent method to obtain a minimum loss function:
wherein L represents a loss function, r represents a reward value, theta and theta' represent network weights,
representing a target Q function value, Q (s, a, theta) representing a predicted Q function value, and gamma representing a discount factor;
and predicting the price change trend of the financial monitoring target by using the trained deep learning network.
The investment prediction model constructed by the invention can realize the price change trend prediction of the financial monitoring target, has accurate prediction and strong robustness, and solves the problems of overlarge Q-learning state space, low convergence speed and the like.
The financial wind control system based on the block chain also provides free financial wind control monitoring for the user node, and the investment index I is larger than the investment index threshold I 0 The investment target screening is used as a financial monitoring target, benefits of most user nodes are guaranteed according to asset allocation of the user nodes, extra expenses are reduced for the user nodes, the user nodes can be attracted to join continuously, and the system health development is facilitated. The monitoring node acquires the comment data of the fusion monitoring target in a preset time period from a preset financial data source platform, preprocesses the comment data, calculates an emotion index E, and if the emotion index E is larger than an emotion index threshold E 0 Then sending a prediction request to a prediction node, and setting an emotion index threshold value E 0 And the emotion index E is continuously monitored according to the comment data, so that the online public sentiment can be timely monitored, and the market fluctuation which may occur is laid out in advance.
Referring to fig. 3, fig. 3 is a flowchart illustrating S101-S104 of a financial wind control method based on a block chain according to an embodiment of the present invention. As shown in fig. 3, the step S10 specifically includes:
s101, the common identification node and the user node respectively send a public key corresponding to the private key to a block chain;
s102, calculating a hash value corresponding to investment data by a user node to serve as a private key, encrypting the investment data through the private key, performing secondary encryption through a public key of a consensus node, and sending encrypted content to a block chain, wherein the investment data comprise an investor ID, an investment target name and a target amount;
s103, the consensus node verifies the investment data after the public key is encrypted according to the private key of the consensus node, and then decrypts the encrypted content through the public key of the user node to obtain the investment data;
s104, the consensus node calculates a hash value corresponding to the investor ID and the name of the investment target, encrypts the investment data through a private key, and then sends the encrypted content to a block chain.
The invention can ensure the safety of the data sharing between the user node and the common identification node through the double encryption sharing of the investment data, only the common identification node can decrypt the investment data of the user node, and the information leakage can be avoided.
Referring to fig. 4, fig. 4 is a flowchart illustrating S301-S303 of a financial wind control method based on a block chain according to an embodiment of the present invention. As shown in fig. 4, in the step S30, the comment data is preprocessed, which includes a first priority process and a second priority process, where the first priority process is as follows:
s301, deleting messy codes, blanks and picture comments;
s302, when the same IP user is in T 1 Number M of identical comments made on same platform in timeThreshold M of not less than the same comment 1 When the comment is deleted, redundant comments are deleted, and only one comment is reserved;
s303, when the same IP user is in T 2 The number of more comments M and more than the number of comments M are published on the same platform in time is more than or equal to the threshold value M of the number of comments 2 When the same IP user is at T 3 The number of the comments M which are published in different platforms within the time is more than or equal to the number of comments threshold M 3 When it is time, delete all its comments, where T 3 ≥T 2 ≥T 1 And M is 2 ≥M 3 。
Wherein, when the same IP user is at T 1 The number M of the same comments published on the same platform in time is more than or equal to the same comment threshold value M 1 When the comment is deleted, redundant comments are deleted, and only one comment is reserved and used for identifying network delay or delay, so that the redundant comments are deleted; when the same IP user is at T 2 More than comment number M is more than or equal to comment number threshold value M when being published on the same platform in time 2 When the same IP user is at T 3 The number of the comments M which are published in different platforms within the time is more than or equal to the number of comments threshold M 3 When it is time to delete all its comments, where T 3 ≥T 2 ≥T 1 And M is 2 ≥M 3 The method is used for identifying the network water army and deleting and filtering the relevant comments. The number of true critiques is obtained by the first priority process.
Referring to fig. 5, fig. 5 is a flowchart illustrating S304-S306 of the financial wind control method based on the blockchain according to the embodiment of the present invention. As shown in fig. 5, the second priority processing includes:
s304, obtaining the comment text, performing word segmentation and stop word and preposition word removing processing to obtain keywords;
s305, screening the keywords according to the screening model F (x), wherein the deletion value is less than the threshold value F 0 Wherein the screening model F (x) is defined as follows:
wherein F (x) represents a filtering function, x represents a serial number of a keyword, q x Representing the number of times of occurrence of the x-th keyword in the comments, c representing the total number of keywords in the comments containing the x-th keyword, N representing the total number of the comments, and N representing the number of the comments of the keyword;
s306, secondary classification is carried out on the screened keywords through a pre-trained good classification model, and positive comments or negative comments are obtained.
It should be noted that the classification model may adopt a common reinforced model in the field for classification training, or may adopt the following classification model for classification training:
acquiring keyword data configured with positive comment or negative comment category labels, dividing the keyword data into a training set and a testing set, and inputting a constructed classification model for training; the construction method for constructing the classification model comprises the following steps:
establishing an optimal reward model:
where E represents an expectation, λ represents a discount factor, λ ∈ [0,1 ]];s
0 Representing the initial state, R representing the reward function, pi(s)
t ) Representing a policy that maps states to operations;
defining the Q function: q πi (s,a)=R(s,a)+λ∑ s* p(s,a,s * )T πi (s) where pi i represents the current strategy to determine the Q value according to the equation, R represents the function, λ represents the refraction factor, p (s, a, s) * ) Indicates that action a transits from state s to s * Transition probability of, T πi Representing the reward obtained by iterating step i;
the iterative update of the new strategy is as follows:
pi (i + 1)(s) = argmaxQ (s, a), defining an epsilon-greedy behavior strategy, and determining the behavior of the current state by adopting the epsilon-greedy behavior strategy, wherein each action is a certain predefined fixed probability
Randomly selecting;
the Q value approaches to the acquisition of an optimal strategy through learning iteration;
and carrying out secondary classification on the screened keywords through a pre-trained good classification model to obtain positive comments or negative comments.
In an embodiment of the present invention, the method further includes:
the monitoring node calculates a divergence index D, if the emotion index E is larger than an emotion index threshold E 0 And the divergence index D is larger than the divergence index threshold value D 0 Sending an assessment request to the expert node, wherein the divergence index D is calculated according to the following formula:
wherein D represents a divergence index, m * Indicating the number of task requests of the user node in the current update period, g 1 The growth rate, p, of positive comments representing the current update period 1 Number of positive comments,/ 1 Indicates positive comment, i 2 Indicates positive comment on the number of persons stepped on, g 2 Growth rate of negative comments, p, representing the current update period 2 Number of negative comments,/ 3 Number of negative comments,/ 4 Representing the number of passive comment steps;
the expert node obtains the prediction result of the prediction node, and adds the expert evaluation suggestion broadcast in the block network.
The invention realizes the solution of the investment problem for investors through the manual intervention of financial patents under the specific condition of large investment divergence by arranging the expert node, the monitoring node calculates the divergence index D, and if the emotion index E is more than the emotion index threshold E 0 And divergence index D > divergence index threshold D 0 And sending an evaluation request to the expert node to provide an evaluation suggestion for the user node, ensuring the utilization of investors and realizing more efficient financial wind control.
As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The invention may be described in the general context of computer-executable instructions, such as program modules, being executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The invention may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media including memory storage devices.
All the embodiments in the invention are described in a progressive manner, and the same and similar parts among the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, as for the system embodiment, since it is substantially similar to the method embodiment, the description is relatively simple, and reference may be made to the partial description of the method embodiment for relevant points.
The above description is only an example of the present invention, and is not intended to limit the present invention. Various modifications and alterations to this invention will become apparent to those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of the claims of the present invention.