CN113450222A - Block chain technology-based method for managing deposit duration bonds - Google Patents

Block chain technology-based method for managing deposit duration bonds Download PDF

Info

Publication number
CN113450222A
CN113450222A CN202110772313.9A CN202110772313A CN113450222A CN 113450222 A CN113450222 A CN 113450222A CN 202110772313 A CN202110772313 A CN 202110772313A CN 113450222 A CN113450222 A CN 113450222A
Authority
CN
China
Prior art keywords
weighted
feature vector
feature vectors
text data
event
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.)
Withdrawn
Application number
CN202110772313.9A
Other languages
Chinese (zh)
Inventor
王仁杰
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hangzhou Xinku Technology Co ltd
Original Assignee
Hangzhou Xinku Technology Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Hangzhou Xinku Technology Co ltd filed Critical Hangzhou Xinku Technology Co ltd
Priority to CN202110772313.9A priority Critical patent/CN113450222A/en
Publication of CN113450222A publication Critical patent/CN113450222A/en
Withdrawn legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
    • G06Q40/04—Trading; Exchange, e.g. stocks, commodities, derivatives or currency exchange
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/27—Replication, distribution or synchronisation of data between databases or within a distributed database system; Distributed database system architectures therefor
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
    • G06F21/60—Protecting data
    • G06F21/64—Protecting data integrity, e.g. using checksums, certificates or signatures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
    • G06Q40/03—Credit; Loans; Processing thereof

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Finance (AREA)
  • Accounting & Taxation (AREA)
  • General Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Computer Security & Cryptography (AREA)
  • General Business, Economics & Management (AREA)
  • Technology Law (AREA)
  • Strategic Management (AREA)
  • Marketing (AREA)
  • Economics (AREA)
  • Development Economics (AREA)
  • Computing Systems (AREA)
  • Evolutionary Computation (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Artificial Intelligence (AREA)
  • Probability & Statistics with Applications (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Bioethics (AREA)
  • General Health & Medical Sciences (AREA)
  • Computer Hardware Design (AREA)
  • Software Systems (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

The method aims at the centralized storage and management functions of distributed data of a block chain and the non-tamper-resistance and real-time performance of the data stored by the block chain, adopts a converter-based semantic understanding model to semantically understand and classify text data of a plurality of current events of the life bond and text data of one or more historical events related to the current events, and accordingly obtains a classification result representing the risk prediction result of the life bond. Therefore, risk prediction is carried out by monitoring the relevant information of the deposit bond, whether the deposit bond right has risk or not is determined, and a subsequent management system of the debt financing tool is perfected.

Description

Block chain technology-based method for managing deposit duration bonds
Technical Field
The present invention relates to the field of blockchain, and more particularly, to a method for managing a lifetime bond based on blockchain technology, a system for managing a lifetime bond based on blockchain technology, and an electronic device.
Background
The block chain is a distributed shared account book and a database, and has the characteristics of decentralization, no tampering, trace retaining in the whole process, traceability, collective maintenance, openness and transparency and the like. The characteristics ensure the honesty and the transparency of the block chain and lay a foundation for creating trust for the block chain. In recent years, as the technology of blockchain matures and develops, various data management technologies based on blockchain technology and applications thereof are developed due to the unique non-alterable characteristic of blockchain.
The technical scheme is that after the deal between banks of China, a supervising center successively releases and revises ' guide for subsequent management work of principal and nominators of non-financial enterprise debt financing tool in the inter-bank bond market, ' guide for emergency management work of non-financial enterprise debt financing tool emergency management work in the inter-bank bond market and ' conference procedure of holders of non-financial enterprise debt financing tool in the inter-bank bond market from 2010, and the like, a scientific work system is established for the subsequent management of the inter-bank bond market, but at present, the inter-bank bond market debt financing tool products have no unified subsequent management information system, and the principal and nominators generally adopt an off-line manual maintenance mode to perform information maintenance and dynamic monitoring on the remaining bond and enterprise information, so that the requirements on labor cost and time cost of the subsequent management are large, and the risk event monitoring and the timely troubleshooting are not favorable.
Therefore, an optimized solution for the management of the lifetime bonds is desired.
Disclosure of Invention
The present application is proposed to solve the above-mentioned technical problems. The embodiment of the application provides a block chain technology-based duration bond management method, a block chain technology-based duration bond management system and an electronic device, aiming at the centralized storage and management functions of distributed data of a block chain and the non-tamper-resistance and real-time performance of the data stored by the block chain, a converter-based semantic understanding model is adopted to carry out semantic understanding and classification on text data of a plurality of current events of the duration bond and text data of one or more historical events related to the text data, and therefore a classification result representing the risk prediction result of the duration bond is obtained. Therefore, risk prediction is carried out by monitoring the relevant information of the deposit bond, whether the deposit bond right has risk or not is determined, and a subsequent management system of the debt financing tool is perfected.
According to one aspect of the application, a method for managing a lifetime bond based on a block chain technology is provided, which comprises the following steps:
obtaining text data of a plurality of current events of a lifetime bond;
respectively passing the text data of the current events through a converter-based semantic understanding model to obtain a plurality of current event feature vectors;
acquiring text data of one or more historical events related to the plurality of current events;
respectively passing the text data of the one or more historical events through the converter-based semantic understanding model to obtain one or more historical event feature vectors;
for each feature vector to be weighted in the plurality of current event feature vectors and the one or more historical event feature vectors, acquiring one or more reference event feature vectors corresponding to each feature vector to be weighted, wherein the reference event feature vectors are obtained by passing template text data of corresponding events through the converter-based semantic understanding model;
for each to-be-weighted feature vector in the plurality of current event feature vectors and the one or more historical event feature vectors, respectively calculating a classification probability weight expression between each to-be-weighted feature vector and the one or more reference event feature vectors corresponding to the to-be-weighted feature vector to obtain a weight value corresponding to each to-be-weighted feature vector, wherein the classification probability weight expression is a quotient of a weighted sum of natural exponent function values raised to the negative of the feature value of each position in the to-be-weighted feature vector and a sum of weighted sums of natural exponent function values raised to the negative of the feature value of each position in the one or more reference event feature vectors;
weighting each feature vector to be weighted by each weight value to obtain a plurality of weighted feature vectors;
after the weighted feature vectors are cascaded, obtaining a classification result through a classifier, wherein the classification result is used for representing a risk prediction result of the bond in the life cycle; and
storing the text data of a plurality of current events of the lifetime bond and the classification result in a storage block of a block chain architecture.
In the above method for managing a lifetime bond based on a blockchain technique, passing the text data of the plurality of current events through a converter-based semantic understanding model respectively to obtain a plurality of current event feature vectors includes: respectively inputting the text data of the current event into a word embedding model to obtain a word vector of the current event; and inputting the word vector of the current event into the converter-based semantic understanding model to obtain the current event feature vector.
In the above method for managing the lifetime bond based on the blockchain technology, the semantic understanding model based on the converter is a Bert model.
In the above method for managing a lifetime bond based on a block chain technique, for each to-be-weighted feature vector in the plurality of current event feature vectors and the one or more historical event feature vectors, respectively calculating a classification probability weight expression between each to-be-weighted feature vector and the one or more reference event feature vectors corresponding to the to-be-weighted feature vector to obtain a weight value corresponding to each to-be-weighted feature vector, the method includes: calculating a classification probability weight expression between each feature vector to be weighted and the one or more reference event feature vectors corresponding to the feature vector to be weighted to obtain a weight value corresponding to each feature vector to be weighted, wherein the formula is as follows: p ═ Σiexp(-xi)/∑i，jexp (-yi), xi represents the eigenvalue of each position of the eigenvector to be weighted, yi represents the eigenvalue of each position of the reference event eigenvector, and j is further summed to represent Σ for the plurality of reference event eigenvectorsiThe values of exp (-yi) are summed.
In the above method for managing a deposit bond based on a block chain technique, the step of cascading the weighted feature vectors and then passing the result through a classifier to obtain a classification result, where the classification result is used to represent a risk prediction result of the deposit bond, includes: after the weighted feature vectors are cascaded, inputting a Softmax classification function to obtain a first probability of belonging to a risk-free classification label and a second probability of belonging to a risk-free classification label; and generating the classification result based on the first probability and the second probability.
In the above method for managing a lifetime bond based on a blockchain technique, the event includes: major matters, bearer meetings, terms of application, credit events, and rating adjustment events.
According to another aspect of the present application, there is provided a block chain technology-based lifetime bond management system, including:
a current text data acquisition unit for acquiring text data of a plurality of current events of the lifetime bond;
a first feature vector generation unit, configured to pass the text data of the multiple current events obtained by the current text data obtaining unit through a converter-based semantic understanding model respectively to obtain multiple current event feature vectors;
a history text data acquisition unit configured to acquire text data of one or more history events related to the plurality of current events;
a second feature vector generation unit, configured to pass the text data of the one or more historical events obtained by the historical text data obtaining unit through the converter-based semantic understanding model respectively to obtain one or more historical event feature vectors;
a reference event feature vector generating unit, configured to obtain, for each to-be-weighted feature vector in the plurality of current event feature vectors obtained by the first feature vector generating unit and the one or more historical event feature vectors obtained by the second feature vector generating unit, one or more reference event feature vectors corresponding to each to-be-weighted feature vector, where the reference event feature vectors are obtained by passing template text data of corresponding events through the converter-based semantic understanding model;
a weight value calculating unit, configured to calculate, for each to-be-weighted feature vector in the plurality of current event feature vectors obtained by the first feature vector generating unit and the one or more historical event feature vectors obtained by the second feature vector generating unit, a classification probability weight expression between each to-be-weighted feature vector and the one or more reference event feature vectors obtained by the reference event feature vector generating unit corresponding thereto, respectively, to obtain a weight value corresponding to each to-be-weighted feature vector, wherein the classification probability weight is expressed as a quotient of a weighted sum of natural exponent function values raised to a power of a negative value of the feature value at each position in the feature vector to be weighted divided by a sum of weighted sums of natural exponent function values raised to a power of a negative value of the feature value at each position in the one or more reference event feature vectors;
a weighted feature vector generating unit, configured to weight each to-be-weighted feature vector by the weight value obtained by each weight value calculating unit to obtain a plurality of weighted feature vectors;
the classification result generating unit is used for cascading the weighted feature vectors obtained by the weighted feature vector generating unit and then obtaining a classification result through a classifier, wherein the classification result is used for representing a risk prediction result of the existing bond; and
a storage unit, configured to store the text data of the multiple current events of the lifetime bond obtained by the current text data obtaining unit and the classification result obtained by the classification result generating unit in a storage block of a block chain architecture.
According to still another aspect of the present application, there is provided an electronic apparatus including: a processor; and a memory having stored therein computer program instructions which, when executed by the processor, cause the processor to perform the method of lifetime bond management based on blockchain techniques as described above.
According to yet another aspect of the present application, there is provided a computer readable medium having stored thereon computer program instructions which, when executed by a processor, cause the processor to perform the method for lifetime bond management based on blockchain techniques as described above.
Compared with the prior art, the block chain technology-based duration bond management method, the block chain technology-based duration bond management system and the electronic device provided by the application have the advantages that the semantic understanding model based on the converter is adopted to semantically understand and classify text data of a plurality of current events of the duration bond and text data of one or more related historical events aiming at the centralized storage and management functions of distributed data of the block chain and the non-tamper-resistance and real-time performance of the data stored by the block chain, so that the classification result representing the risk prediction result of the duration bond is obtained. Therefore, risk prediction is carried out by monitoring the relevant information of the deposit bond, whether the deposit bond right has risk or not is determined, and a subsequent management system of the debt financing tool is perfected.
Drawings
The above and other objects, features and advantages of the present application will become more apparent by describing in more detail embodiments of the present application with reference to the attached drawings. The accompanying drawings are included to provide a further understanding of the embodiments of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the principles of the application. In the drawings, like reference numbers generally represent like parts or steps.
Fig. 1 is a schematic diagram of a block chain-based lifetime bond database according to an embodiment of the present application;
fig. 2 is an application scenario diagram of a method for managing a lifetime bond based on a blockchain technique according to an embodiment of the present application;
fig. 3 is a flowchart of a method for managing lifetime bonds based on a block chain technique according to an embodiment of the present application;
fig. 4 is a schematic diagram of a system architecture of a method for managing lifetime bonds based on a block chain technique according to an embodiment of the present application;
fig. 5 is a flowchart of passing text data of a plurality of current events through a converter-based semantic understanding model to obtain a plurality of current event feature vectors, respectively, in a method for managing a lifetime bond based on a blockchain technique according to an embodiment of the present application;
fig. 6 is a flowchart illustrating a classification result obtained by cascading the weighted feature vectors through a classifier in the method for managing a lifetime bond based on a block chain technique according to an embodiment of the present application;
fig. 7 is a block diagram of a block chain technology based on a life bond management system according to an embodiment of the present application;
fig. 8 is a block diagram of a first feature vector generation unit in a block chain technology-based lifetime bond management system according to an embodiment of the present application;
fig. 9 is a block diagram of a classification result generation unit in the block chain technology-based lifetime bond management system according to an embodiment of the present application;
fig. 10 is a block diagram of an electronic device according to an embodiment of the application.
Detailed Description
Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be understood that the described embodiments are only some embodiments of the present application and not all embodiments of the present application, and that the present application is not limited by the example embodiments described herein.
Block chain architecture overview
Fig. 1 illustrates an architectural diagram of a blockchain-based lifetime bond database according to an embodiment of the present application. As shown in fig. 1, the block chain-based lifetime bond database according to the embodiment of the present application employs a typical block chain architecture, and related information of the lifetime bond, such as important matters, holder meetings, insurance terms, credit events, rating adjustment events P1, P2, …, Pn, is stored in each storage block B1, B2, …, Bn constructed in a block chain. Of course, those skilled in the art will appreciate that different types of life bonds may also be stored in separate blocks, for example, one block dedicated to storing life bonds for credit events and another block dedicated to storing life bonds for rating adjustment events.
According to a typical blockchain storage architecture, each block B1, B2, …, Bn includes pointers H1, H2, …, Hn and data portions D1, D2, …, Dn. The pointers H1, H2, …, Hn may be various types of hash pointers, such as SHA-256 hash functions commonly used in blockchain storage architectures, that point to the last chunk.
In the embodiment of the present application, the value of the hash pointer of the next chunk is based on the value of the hash pointer of the previous chunk and the hash function value of the data portion, for example, H2 ═ H1 × H (D1), and H (D1) represents the hash function value of the data portion D1. The value of the hash pointer for the first chunk may be a random value. In this way, any modification to the portion of data within a block will react on the value of the hash pointer of the next block and further change the values of the hash pointers of all subsequent blocks, making modifications to the portion of data virtually impossible.
Also, in each data portion D1, D2, …, Dn, the hash function value for that data portion may be based on a hash function value generated separately for each of the lifetime bonds in that data portion. For example, all of the lifetime bonds in the data portion may be stored in a hash pointer based data structure of a merkel tree, thereby facilitating tracing back to a particular lifetime bond via the hash pointer and establishing appropriate membership between the respective lifetime bonds.
Here, it can be understood by those skilled in the art that the blockchain-based lifetime bond database according to the embodiment of the present application may adopt any general blockchain architecture, and the embodiment of the present application is not intended to limit the specific implementation of the blockchain architecture.
Moreover, in the embodiment of the present application, the block chain preferably adopts a private chain or a federation chain, so as to facilitate distributed storage management of the storage duration bond database in the inter-bank bond market, and accordingly, each storage block for storing the relevant information of the storage duration bond can be configured in advance without being generated based on the consensus algorithm, so that consumption of computing resources caused by the consensus algorithm can be avoided.
That is to say, the blockchain architecture of the blockchain-based remaining-period bond database according to the embodiment of the present application focuses on storage management of relevant information of remaining-period bonds, and does not relate to a value transfer function based on blockchains similar to electronic money, so that the blockchain architecture can be configured in advance in a cloud by a management department in an inter-bank bond market, accessed from a terminal by each technical department, uploaded with relevant information of remaining-period bonds, and uniformly stored and managed in the cloud. Therefore, since the technical departments are likely to be distributed in different geographic locations, the application of the blockchain architecture can conveniently realize the distributed storage of the lifetime bonds.
On the other hand, each block in the block chain architecture according to the embodiment of the present application may also be associated with a block of the public chain, so that each block has time stamp information corresponding to the associated block of the public chain. Thus, when information requiring a time attribute, such as the uploading time of a lifetime bond, needs to be recorded to determine whether the lifetime bond is an early version, the time sequence attribute of each block in the block chain can be utilized.
Overview of a scene
As mentioned above, currently, there is no unified subsequent management information system for inter-bank market debt financing tool products, and the main underwriters generally adopt an offline manual maintenance mode to perform information maintenance and dynamic monitoring on the deposit bond and the enterprise information, which has a large requirement on the labor cost and time cost of subsequent management, and is not beneficial to timely monitoring and troubleshooting of risk events.
Based on this, the inventor of the present application adopts a block chain based storage architecture for information storage and management of relevant information monitoring of a lifetime bond, including major issue monitoring, holder meeting monitoring, insurance clause monitoring, credit event monitoring, and rating adjustment event monitoring, which is because the above information has a high requirement for authenticity and requires non-tamper-ability of block chain technology to ensure data authenticity. In addition, as the block chain is a storage framework with a time stamp, the timeliness of the information can be ensured.
In addition, aiming at the above events, namely important matters, holder meetings, insurance application terms, credit events and rating adjustment events, the inventor further predicts the risk event of the credit right of the deposit term through the above events, namely, in the case of the above events, determines whether the deposit term is at risk or not through an artificial intelligence mode, thereby perfecting the subsequent management system of the debt financing tool.
Currently, a text classification task based on a semantic understanding model has been greatly developed in the field of artificial intelligence, and particularly, a semantic understanding model based on a converter (transformer) such as Bert is very suitable for semantic understanding and classification in some vertical fields because it introduces an entity word vector (entity) in addition to a general word vector (token). Therefore, in the technical solution of the present application, a converter-based semantic understanding model is used to predict a risk in the event.
In the technical solution of the present application, due to the relevance between events, prediction is not performed by events of a certain time, but prediction is performed by accumulated events, that is, after text data of an event to be considered is obtained and converted into an input vector, and further converted into a feature vector by a semantic understanding model based on a converter, classification is performed based on all feature vectors of all events that have occurred historically. Therefore, in the technical solution of the present application, it is necessary to store historical event feature vectors of events that have occurred historically.
That is, in the technical solution of the present application, text data of a current event is first acquired and converted into a series of word vectors through a word embedding model, and then a current event feature vector is acquired through a converter-based semantic understanding model, and at the same time, one or more historical event feature vectors of stored historical events are acquired. Then, in order to determine a weighted relationship between the current event feature vector and the historical event feature vector, the inventors of the present application employed a classification probability weight expression based on template data, i.e., a reference event feature vector. That is, for each feature vector to be weighted in the current event feature vector and the historical event feature vector, the event type to which it belongs, such as the important item, the holder meeting, the insurance clause, the credit event and the rating adjustment event as described above, is first determined, and then the corresponding reference event feature vector under the event type is determined. Here, the reference event feature vector may be a reference event feature vector obtained by subjecting template text data of the above event to a converter-based semantic understanding model, and since a certain type generally corresponds to a plurality of template texts, also corresponds to a plurality of reference event feature vectors.
Next, a classification probability weight expression between the feature vector to be weighted and a plurality of reference event feature vectors is calculated, i.e. p ═ Σiexp(-xi)/∑i，jexp (-yi), where xi refers to the eigenvalue of each position of the eigenvector to be weighted and yi refers to the eigenvalue of each position of the reference event eigenvector, and j is further summed to represent Σ for the plurality of reference event eigenvectorsiThe values of exp (-yi) are summed. In this way, the weight of each feature vector to be weighted can be calculated, and the weight is substantially used for indicating the degree of attention required by the event occurring at this time in the event of the same type.
Then, weighting each feature vector to be weighted according to the calculated weight, cascading the weighted feature vectors to obtain a classification feature vector, and further obtaining a classification result through a classifier, wherein the classification result can represent a risk prediction result. In addition, the risk prediction result and the text data of the current event should be stored in the storage block of the block chain architecture, so as to realize the risk management of the lifetime creditor.
Based on this, the present application provides a method for managing a lifetime bond based on a block chain technology, which includes: obtaining text data of a plurality of current events of a lifetime bond; respectively passing the text data of the current events through a converter-based semantic understanding model to obtain a plurality of current event feature vectors; acquiring text data of one or more historical events related to the plurality of current events; respectively passing the text data of the one or more historical events through the converter-based semantic understanding model to obtain one or more historical event feature vectors; for each feature vector to be weighted in the plurality of current event feature vectors and the one or more historical event feature vectors, acquiring one or more reference event feature vectors corresponding to each feature vector to be weighted, wherein the reference event feature vectors are obtained by passing template text data of corresponding events through the converter-based semantic understanding model; for each to-be-weighted feature vector in the plurality of current event feature vectors and the one or more historical event feature vectors, respectively calculating a classification probability weight expression between each to-be-weighted feature vector and the one or more reference event feature vectors corresponding to the to-be-weighted feature vector to obtain a weight value corresponding to each to-be-weighted feature vector, wherein the classification probability weight expression is a quotient of a weighted sum of natural exponent function values raised to the negative of the feature value of each position in the to-be-weighted feature vector and a sum of weighted sums of natural exponent function values raised to the negative of the feature value of each position in the one or more reference event feature vectors; weighting each feature vector to be weighted by each weight value to obtain a plurality of weighted feature vectors; after the weighted feature vectors are cascaded, obtaining a classification result through a classifier, wherein the classification result is used for representing a risk prediction result of the bond in the life cycle; and storing the text data of a plurality of current events of the lifetime bonds and the classification result in a storage block of a block chain architecture.
Fig. 2 illustrates an application scenario of a method for managing lifetime bonds based on a blockchain technology according to an embodiment of the present application. As shown in fig. 2, in this application scenario, first, text data of a plurality of current events of a lifetime bond and text data of one or more historical events related to the plurality of current events are acquired through a terminal device (e.g., D as illustrated in fig. 2); then, the text data of the current event and the text data of the historical event are input into a server (e.g., a cloud server S as illustrated in fig. 2) deployed with a life bond management algorithm based on the blockchain technology, wherein the server can process the text data of the current event and the text data of the historical event based on the life bond management algorithm of the blockchain technology to generate a risk prediction result representing the life bond. Then, the text data of the plurality of current events of the lifetime bond and the classification result are stored in a storage block of a block chain architecture (e.g., block T as illustrated in fig. 2).
Having described the general principles of the present application, various non-limiting embodiments of the present application will now be described with reference to the accompanying drawings.
Exemplary method
Fig. 3 illustrates a flow chart of a method of life bond management based on blockchain technology. As shown in fig. 3, a method for managing a lifetime bond based on a block chain technology according to an embodiment of the present application includes: s110, acquiring text data of a plurality of current events of the bond in the life cycle; s120, respectively passing the text data of the current events through a semantic understanding model based on a converter to obtain a plurality of current event feature vectors; s130, acquiring text data of one or more historical events related to the current events; s140, respectively passing the text data of the one or more historical events through the converter-based semantic understanding model to obtain one or more historical event feature vectors; s150, for each feature vector to be weighted in the plurality of current event feature vectors and the one or more historical event feature vectors, obtaining one or more reference event feature vectors corresponding to each feature vector to be weighted, wherein the reference event feature vectors are obtained by the converter-based semantic understanding model for template text data of corresponding events; s160, for each to-be-weighted feature vector in the plurality of current event feature vectors and the one or more historical event feature vectors, respectively calculating a classification probability weight expression between each to-be-weighted feature vector and the one or more reference event feature vectors corresponding thereto to obtain a weight value corresponding to each to-be-weighted feature vector, wherein the classification probability weight expression is a quotient of a weighted sum of natural exponent function values raised to a power of a negative value of the feature value at each position in the to-be-weighted feature vector and a sum of weighted sums of natural exponent function values raised to a power of a negative value of the feature value at each position in the one or more reference event feature vectors; s170, weighting each feature vector to be weighted by each weight value to obtain a plurality of weighted feature vectors; s180, cascading the weighted feature vectors and then passing the weighted feature vectors through a classifier to obtain a classification result, wherein the classification result is used for representing a risk prediction result of the existing bond; and S190, storing the text data of the current events of the lifetime bonds and the classification result in a storage block of a block chain architecture.
Fig. 4 is a schematic diagram illustrating an architecture of a method for managing lifetime bonds based on a block chain technique according to an embodiment of the present application. As shown IN fig. 4, IN the network architecture of the method for managing a lifetime bond based on the blockchain technology, first, text data of a plurality of current events of the lifetime bond (for example, IN11 to IN1n as illustrated IN fig. 4) are acquired; then, passing the text data of the plurality of current events through a converter-based semantic understanding model (e.g., SUM as illustrated in fig. 4) to obtain a plurality of current event feature vectors (e.g., V11 to V1n as illustrated in fig. 4), respectively; next, acquiring text data of one or more historical events related to the plurality of current events (e.g., IN 21-IN 2n as illustrated IN fig. 4); then, passing the text data of the one or more historical events through the converter-based semantic understanding model (e.g., SUM as illustrated in fig. 4) to obtain one or more historical event feature vectors (e.g., V21-V2 n as illustrated in fig. 4), respectively; then, for each feature vector to be weighted in the plurality of current event feature vectors and the one or more historical event feature vectors, acquiring one or more reference event feature vectors (e.g., Vr11 to Vr2n as illustrated in fig. 4) corresponding to each feature vector to be weighted, wherein the reference event feature vectors are obtained by the converter-based semantic understanding model for template text data of the corresponding event; then, for each feature vector to be weighted in the plurality of current event feature vectors and the one or more historical event feature vectors, respectively calculating a classification probability weight expression between each feature vector to be weighted and the one or more reference event feature vectors corresponding to the feature vector to be weighted to obtain a weight value corresponding to each feature vector to be weighted (e.g., as illustrated in fig. 4, W11 to W2 n); then, weighting each of the feature vectors to be weighted with each of the weighting values to obtain a plurality of weighted feature vectors (e.g., Vw 11-Vw 2n as illustrated in fig. 4); then, cascading the weighted feature vectors and passing the concatenated feature vectors through a classifier (for example, a circle S as illustrated in fig. 4) to obtain a classification result, wherein the classification result is used for representing a risk prediction result of the lifetime bond; then, the text data of the plurality of current events of the lifetime bond and the classification result are stored in a storage block of a block chain architecture (e.g., T as illustrated in fig. 4).
In step S110, text data of a plurality of current events of the lifetime bond is acquired.
Specifically, in the embodiment of the present application, the event includes: major matters, bearer meetings, terms of application, credit events, and rating adjustment events. It should be understood that in the present application, the risk event of the deposit right is predicted through the above events, that is, in the case of the above event, whether the deposit right is at risk is determined through an artificial intelligence manner, so as to perfect the subsequent management system of the debt financing tool.
In step S120, the text data of the plurality of current events are respectively passed through a converter-based semantic understanding model to obtain a plurality of current event feature vectors.
Specifically, in this embodiment of the present application, the process of passing the text data of the plurality of current events through a converter-based semantic understanding model to obtain a plurality of current event feature vectors includes: firstly, the text data of the current event are respectively input into a word embedding model to obtain a word vector of the current event. It should be understood that text is a very important type of unstructured data, and text can be converted into structured data through a bag-of-words model, TF-IDF, a topic model and a word embedding model, i.e., text data is represented in a vector form. Here, text data of the current event in text form is converted into a Word vector of the current event in a Word embedding model such as Word2Vec or the like. Then, the word vector of the current event is input into the converter-based semantic understanding model to obtain the current event feature vector. That is, semantic features in the word vector of the current event are obtained through a semantic understanding model to obtain a current event feature vector.
In particular, in the embodiment of the present application, the semantic understanding model based on the converter is a Bert model. Those skilled in the art will appreciate that, at present, the task of text classification based on semantic understanding model has been greatly developed in the field of artificial intelligence, and especially, the semantic understanding model based on converter, such as Bert, is very suitable for semantic understanding and classification in some vertical fields because it introduces entity word vectors in addition to general word vectors.
Fig. 5 illustrates a flowchart of passing text data of the current events through a converter-based semantic understanding model to obtain a plurality of current event feature vectors, respectively, in a method for managing a lifetime bond based on a blockchain technology according to an embodiment of the present application. As shown in fig. 5, in the embodiment of the present application, passing the text data of the plurality of current events through a converter-based semantic understanding model to obtain a plurality of current event feature vectors respectively includes: s210, respectively inputting the text data of the current event into a word embedding model to obtain a word vector of the current event; and S220, inputting the word vector of the current event into the converter-based semantic understanding model to obtain the current event feature vector.
In step S130, text data of one or more historical events related to the plurality of current events is acquired. It should be understood that due to the relevance between events, prediction is not performed through a certain event, but through accumulated events, and therefore, in the technical solution of the present application, text data of one or more historical events related to the multiple current events needs to be acquired.
In step S140, the text data of the one or more historical events are respectively passed through the converter-based semantic understanding model to obtain one or more historical event feature vectors. That is, the text data of the historical events are respectively input into a word embedding model to obtain word vectors of the historical events. Then, the word vector of the historical event is input into the converter-based semantic understanding model to obtain the historical event feature vector. In particular, in the embodiment of the present application, the semantic understanding model based on the converter is a Bert model.
In step S150, for each to-be-weighted feature vector in the plurality of current event feature vectors and the one or more historical event feature vectors, one or more reference event feature vectors corresponding to each to-be-weighted feature vector are obtained, where the reference event feature vectors are obtained by passing template text data of a corresponding event through the converter-based semantic understanding model.
It should be appreciated that in order to determine the weighted relationship between the current event feature vector and the historical event feature vector, the inventors of the present application employ a classification probability weight expression based on template data, i.e., a reference event feature vector. That is, for each feature vector to be weighted in the current event feature vector and the historical event feature vector, the event type to which it belongs, such as the important item, the holder meeting, the insurance clause, the credit event and the rating adjustment event as described above, is first determined, and then the corresponding reference event feature vector under the event type is determined. Here, the reference event feature vector may be a reference event feature vector obtained by subjecting template text data of the above event to a converter-based semantic understanding model, and since a certain type generally corresponds to a plurality of template texts, also corresponds to a plurality of reference event feature vectors.
In step S160, for each to-be-weighted feature vector of the plurality of current event feature vectors and the one or more historical event feature vectors, respectively calculating a classification probability weight expression between each to-be-weighted feature vector and the one or more reference event feature vectors corresponding thereto to obtain a weight value corresponding to each to-be-weighted feature vector, wherein the classification probability weight expression is a quotient of a weighted sum of natural exponent function values raised to a negative value of the feature value of each position in the to-be-weighted feature vector and a weighted sum of natural exponent function values raised to a negative value of the feature value of each position in the one or more reference event feature vectors.
Specifically, in this embodiment of the present application, for each to-be-weighted feature vector in the plurality of current event feature vectors and the one or more historical event feature vectors, a process of calculating a classification probability weight expression between each to-be-weighted feature vector and the one or more reference event feature vectors corresponding thereto to obtain a weight value corresponding to each to-be-weighted feature vector includes: calculating a classification probability weight expression between each feature vector to be weighted and the one or more reference event feature vectors corresponding to the feature vector to be weighted to obtain a weight value corresponding to each feature vector to be weighted, wherein the formula is as follows: p ═ Σiexp(-xi)/∑i，jexp (-yi), xi represents the eigenvalue of each position of the eigenvector to be weighted, yi represents the eigenvalue of each position of the reference event eigenvector, and j is further summed to represent Σ for the plurality of reference event eigenvectorsiThe values of exp (-yi) are summed. In this way, the weight of each feature vector to be weighted can be calculated, and the weight is substantially used for indicating the degree of attention required by the event occurring at this time in the event of the same type.
In step S170, each of the feature vectors to be weighted is weighted by each of the weighting values to obtain a plurality of weighted feature vectors. It should be understood that, by weighting each of the feature vectors to be weighted, the obtained weighted feature vectors can better reflect the overall features.
In step S180, the weighted feature vectors are cascaded and then passed through a classifier to obtain a classification result, where the classification result is used to represent a risk prediction result of a lifetime bond.
Specifically, in the embodiment of the present application, the process of cascading the weighted feature vectors and then obtaining the classification result by the classifier includes: first, the weighted feature vectors are concatenated and input into a Softmax classification function to obtain a first probability of being attributed to a risk-free class label and a second probability of being attributed to a risk-free class label. That is, after the weighted feature vectors are spliced into the classification feature vectors, the classification feature vectors are input into a Softmax classification function to obtain probability values corresponding to different classification labels. Here, the classification label may also be a multi-label, for example, three classification labels that can be classified into high risk, low risk and no risk, etc. Then, based on the first probability and the second probability, the classification result is generated. That is, the greater of the first probability and the second probability is determined as the classification result.
Fig. 6 is a flowchart illustrating that the plurality of weighted feature vectors are cascaded and then passed through a classifier to obtain a classification result in a method for managing a lifetime bond based on a blockchain technique according to an embodiment of the present application. As shown in fig. 6, in the embodiment of the present application, the cascading the weighted feature vectors and then passing through the classifier to obtain the classification result includes: s310, after the weighted feature vectors are cascaded, inputting a Softmax classification function to obtain a first probability of belonging to a risk-free classification label and a second probability of belonging to a risk-free classification label; and S320, generating the classification result based on the first probability and the second probability.
In step S190, the text data of the plurality of current events of the lifetime bond and the classification result are stored in a storage block of a block chain architecture. That is, the risk prediction result and the text data of the current event are stored in the storage block of the blockchain architecture, so as to realize the risk management of the lifetime creditor. It should be appreciated that storing text data and the classification results for a plurality of current events via a blockchain may take advantage of the non-tamper-ability of blockchain techniques to ensure data authenticity. In addition, as the block chain is a storage framework with a time stamp, the timeliness of the information can be ensured.
In summary, a method for managing a lifetime bond based on a blockchain technology is disclosed, which employs a converter-based semantic understanding model to semantically understand and classify text data of a plurality of current events of the lifetime bond and text data of one or more historical events related thereto, aiming at the centralized storage and management functions of distributed data of blockchains and the non-tamper-resistance and real-time performance of data stored in blockchains, so as to obtain a classification result representing a risk prediction result of the lifetime bond. Therefore, risk prediction is carried out by monitoring the relevant information of the deposit bond, whether the deposit bond right has risk or not is determined, and a subsequent management system of the debt financing tool is perfected.
Exemplary System
Fig. 7 illustrates a block diagram of a block chain technology based on a life cycle bond management system according to an embodiment of the application. As shown in fig. 7, a system 700 for managing lifetime bonds based on a block chain technique according to an embodiment of the present application includes: a current text data acquisition unit 710 for acquiring text data of a plurality of current events of the lifetime bond; a first feature vector generating unit 720, configured to pass the text data of the multiple current events obtained by the current text data obtaining unit 710 through a converter-based semantic understanding model respectively to obtain multiple current event feature vectors; a historical text data obtaining unit 730, configured to obtain text data of one or more historical events related to the plurality of current events; a second feature vector generating unit 740, configured to pass the text data of the one or more historical events obtained by the historical text data obtaining unit 730 through the converter-based semantic understanding model respectively to obtain one or more historical event feature vectors; a reference event feature vector generating unit 750, configured to obtain, for each to-be-weighted feature vector in the plurality of current event feature vectors obtained by the first feature vector generating unit 720 and the one or more historical event feature vectors obtained by the second feature vector generating unit 740, one or more reference event feature vectors corresponding to each to-be-weighted feature vector, where the reference event feature vectors are obtained by passing template text data of corresponding events through the converter-based semantic understanding model; a weight value calculating unit 760, configured to calculate, for each to-be-weighted feature vector in the plurality of current event feature vectors obtained by the first feature vector generating unit 720 and the one or more historical event feature vectors obtained by the second feature vector generating unit 740, a classification probability weight expression between each to-be-weighted feature vector and the one or more reference event feature vectors obtained by the reference event feature vector generating unit 750 corresponding thereto respectively to obtain a weight value corresponding to each to-be-weighted feature vector, wherein the classification probability weight is expressed as a quotient of a weighted sum of natural exponent function values raised to a power of a negative value of the feature value at each position in the feature vector to be weighted divided by a sum of weighted sums of natural exponent function values raised to a power of a negative value of the feature value at each position in the one or more reference event feature vectors; a weighted feature vector generation unit 770, configured to weight each feature vector to be weighted by the weight value obtained by each weight value calculation unit 760 to obtain a plurality of weighted feature vectors; a classification result generating unit 780, configured to cascade the plurality of weighted feature vectors obtained by the weighted feature vector generating unit 770, and then pass through a classifier to obtain a classification result, where the classification result is used to indicate a risk prediction result of a lifetime bond; and a storage unit 790, configured to store the text data of the plurality of current events of the lifetime bonds obtained by the current text data obtaining unit 710 and the classification result obtained by the classification result generating unit 780 in a storage block of a block chain architecture.
In one example, in the lifetime bond management system 700 described above, as shown in fig. 8, the first feature vector generation unit 720 includes: a word vector generating subunit 721, configured to input the text data of the current event into word embedding models respectively to obtain word vectors of the current event; and a feature vector generation subunit 722 configured to input the word vector of the current event obtained by the word vector generation subunit 721 into the converter-based semantic understanding model to obtain the current event feature vector.
In one example, in the lifetime bond management system 700 described above, the converter-based semantic understanding model is a Bert model.
In one example, in the lifetime bond management system 700, the weight value calculating unit 760 is further configured to: calculating a classification probability weight expression between each feature vector to be weighted and the one or more reference event feature vectors corresponding to the feature vector to be weighted to obtain a weight value corresponding to each feature vector to be weighted, wherein the formula is as follows: p ═ Σiexp(-xi)/∑i，jexp (-yi), xi represents the eigenvalue of each position of the eigenvector to be weighted, yi represents the eigenvalue of each position of the reference event eigenvector, and j is further summed to represent Σ for the plurality of reference event eigenvectorsiThe values of exp (-yi) are summed.
In one example, in the lifetime bond management system 700 described above, as shown in fig. 9, the classification result generation unit 780 includes: a probability generating subunit 781, configured to input the plurality of weighted feature vectors to a Softmax classification function after being concatenated, so as to obtain a first probability of belonging to a risk-free classification tag and a second probability of belonging to a risk-free classification tag; and a classification subunit 782, configured to generate the classification result based on the first probability and the second probability obtained by the probability generation subunit 781.
In one example, in the above-described lifetime bond management system 700, the event includes: major matters, bearer meetings, terms of application, credit events, and rating adjustment events.
Here, it can be understood by those skilled in the art that the detailed functions and operations of the respective units and modules in the above-described renewal bond management system 700 have been described in detail in the above description of the renewal bond management method based on the block chain technology with reference to fig. 1 to 6, and thus, the repeated description thereof will be omitted.
As described above, the present invention can be implemented in various terminal devices, such as a server for use in the deposit bond management, and the like. In one example, the lifetime bond management system 700 according to embodiments of the present application may be integrated into the terminal device as one software module and/or hardware module. For example, the lifetime bond management system 700 may be a software module in the operating system of the terminal device, or may be an application developed for the terminal device; of course, the lifetime bond management system 700 may also be one of many hardware modules of the terminal device.
Alternatively, in another example, the deposit bond management system 700 and the terminal device may be separate devices, and the deposit bond management system 700 may be connected to the terminal device through a wired and/or wireless network and transmit the interactive information according to an agreed data format.
Exemplary electronic device
Next, an electronic apparatus according to an embodiment of the present application is described with reference to fig. 10. As shown in fig. 10, the electronic device 10 includes one or more processors 11 and a memory 12. The processor 11 may be a Central Processing Unit (CPU) or other form of processing unit having data processing capabilities and/or instruction execution capabilities, and may control other components in the electronic device 10 to perform desired functions.
Memory 12 may include one or more computer program products that may include various forms of computer-readable storage media, such as volatile memory and/or non-volatile memory. The volatile memory may include, for example, Random Access Memory (RAM), cache memory (cache), and/or the like. The non-volatile memory may include, for example, Read Only Memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium and executed by the processor 11 to implement the functions of the method for blockchain technology-based lifetime bond management of the various embodiments of the present application described above and/or other desired functions. Various content such as reference event feature vectors, weighted feature vectors, and the like may also be stored in the computer-readable storage medium.
In one example, the electronic device 10 may further include: an input system 13 and an output system 14, which are interconnected by a bus system and/or other form of connection mechanism (not shown).
The input system 13 may comprise, for example, a keyboard, a mouse, etc.
The output system 14 may output various information including classification results and the like to the outside. The output system 14 may include, for example, a display, speakers, a printer, and a communication network and its connected remote output devices, among others.
Of course, for simplicity, only some of the components of the electronic device 10 relevant to the present application are shown in fig. 10, and components such as buses, input/output interfaces, and the like are omitted. In addition, the electronic device 10 may include any other suitable components depending on the particular application.

Claims (10)

1. A method for managing a deposit duration bond based on a block chain technology is characterized by comprising the following steps:
obtaining text data of a plurality of current events of a lifetime bond;
respectively passing the text data of the current events through a converter-based semantic understanding model to obtain a plurality of current event feature vectors;
acquiring text data of one or more historical events related to the plurality of current events;
respectively passing the text data of the one or more historical events through the converter-based semantic understanding model to obtain one or more historical event feature vectors;
for each feature vector to be weighted in the plurality of current event feature vectors and the one or more historical event feature vectors, acquiring one or more reference event feature vectors corresponding to each feature vector to be weighted, wherein the reference event feature vectors are obtained by passing template text data of corresponding events through the converter-based semantic understanding model;
for each to-be-weighted feature vector in the plurality of current event feature vectors and the one or more historical event feature vectors, respectively calculating a classification probability weight expression between each to-be-weighted feature vector and the one or more reference event feature vectors corresponding to the to-be-weighted feature vector to obtain a weight value corresponding to each to-be-weighted feature vector, wherein the classification probability weight expression is a quotient of a weighted sum of natural exponent function values raised to the negative of the feature value of each position in the to-be-weighted feature vector and a sum of weighted sums of natural exponent function values raised to the negative of the feature value of each position in the one or more reference event feature vectors;
weighting each feature vector to be weighted by each weight value to obtain a plurality of weighted feature vectors;
after the weighted feature vectors are cascaded, obtaining a classification result through a classifier, wherein the classification result is used for representing a risk prediction result of the bond in the life cycle; and
storing the text data of a plurality of current events of the lifetime bond and the classification result in a storage block of a block chain architecture.
2. The method for managing the lifetime bonds based on the blockchain technology as claimed in claim 1, wherein the step of passing the text data of the plurality of current events through a converter-based semantic understanding model to obtain a plurality of current event feature vectors respectively comprises:
respectively inputting the text data of the current event into a word embedding model to obtain a word vector of the current event; and
inputting the word vector of the current event into the converter-based semantic understanding model to obtain the current event feature vector.
3. The method for life bond management based on blockchain technology of claim 2, wherein the converter-based semantic understanding model is a Bert model.
4. The method for managing the lifetime bonds based on the block chain technique according to claim 1, wherein for each feature vector to be weighted in the plurality of current event feature vectors and the one or more historical event feature vectors, respectively calculating a classification probability weight expression between each feature vector to be weighted and the one or more reference event feature vectors corresponding to the feature vector to be weighted to obtain a weight value corresponding to each feature vector to be weighted comprises:
calculating a classification probability weight expression between each feature vector to be weighted and the one or more reference event feature vectors corresponding to the feature vector to be weighted to obtain a weight value corresponding to each feature vector to be weighted, wherein the formula is as follows: p ═ Σiexp(-xi)/∑i，jexp (-yi), xi represents the eigenvalue of each position of the eigenvector to be weighted, yi represents the eigenvalue of each position of the reference event eigenvector, and j is further summed to represent Σ for the plurality of reference event eigenvectorsiThe values of exp (-yi) are summed.
5. The method for managing the existing bond based on the block chain technology as claimed in claim 1, wherein the step of cascading the weighted feature vectors to obtain a classification result through a classifier, wherein the classification result is used for representing a risk prediction result of the existing bond, comprises:
after the weighted feature vectors are cascaded, inputting a Softmax classification function to obtain a first probability of belonging to a risk-free classification label and a second probability of belonging to a risk-free classification label; and
generating the classification result based on the first probability and the second probability.
6. The method for life cycle bond management based on block chain technology of claim 1, wherein the event comprises: major matters, bearer meetings, terms of application, credit events, and rating adjustment events.
7. A system for managing a life-cycle bond based on a block chain technique, comprising:
a current text data acquisition unit for acquiring text data of a plurality of current events of the lifetime bond;
a first feature vector generation unit, configured to pass the text data of the multiple current events obtained by the current text data obtaining unit through a converter-based semantic understanding model respectively to obtain multiple current event feature vectors;
a history text data acquisition unit configured to acquire text data of one or more history events related to the plurality of current events;
a second feature vector generation unit, configured to pass the text data of the one or more historical events obtained by the historical text data obtaining unit through the converter-based semantic understanding model respectively to obtain one or more historical event feature vectors;
a reference event feature vector generating unit, configured to obtain, for each to-be-weighted feature vector in the plurality of current event feature vectors obtained by the first feature vector generating unit and the one or more historical event feature vectors obtained by the second feature vector generating unit, one or more reference event feature vectors corresponding to each to-be-weighted feature vector, where the reference event feature vectors are obtained by passing template text data of corresponding events through the converter-based semantic understanding model;
a weight value calculating unit, configured to calculate, for each to-be-weighted feature vector in the plurality of current event feature vectors obtained by the first feature vector generating unit and the one or more historical event feature vectors obtained by the second feature vector generating unit, a classification probability weight expression between each to-be-weighted feature vector and the one or more reference event feature vectors obtained by the reference event feature vector generating unit corresponding thereto, respectively, to obtain a weight value corresponding to each to-be-weighted feature vector, wherein the classification probability weight is expressed as a quotient of a weighted sum of natural exponent function values raised to a power of a negative value of the feature value at each position in the feature vector to be weighted divided by a sum of weighted sums of natural exponent function values raised to a power of a negative value of the feature value at each position in the one or more reference event feature vectors;
a weighted feature vector generating unit, configured to weight each to-be-weighted feature vector by the weight value obtained by each weight value calculating unit to obtain a plurality of weighted feature vectors;
the classification result generating unit is used for cascading the weighted feature vectors obtained by the weighted feature vector generating unit and then obtaining a classification result through a classifier, wherein the classification result is used for representing a risk prediction result of the existing bond; and
a storage unit, configured to store the text data of the multiple current events of the lifetime bond obtained by the current text data obtaining unit and the classification result obtained by the classification result generating unit in a storage block of a block chain architecture.
8. The system for managing the lifetime bonds based on the block chain technique as claimed in claim 7, wherein said first feature vector generating unit comprises:
the word vector generating subunit is used for respectively inputting the text data of the current event into a word embedding model to obtain a word vector of the current event; and
a feature vector generating subunit, configured to input the word vector of the current event obtained by the word vector generating subunit into the converter-based semantic understanding model to obtain the current event feature vector.
9. The system for managing the existing bond based on the block chain technology as claimed in claim 7, wherein the classification result generating unit comprises:
a probability generating subunit, configured to input the plurality of weighted feature vectors into a Softmax classification function after concatenation to obtain a first probability of belonging to a risk-free classification label and a second probability of belonging to a risk-free classification label; and
and the classification subunit is used for generating the classification result based on the first probability and the second probability obtained by the probability generation subunit.
10. An electronic device, comprising:
a processor; and
a memory having stored therein computer program instructions which, when executed by the processor, cause the processor to perform the method of block chain technology based life bond management according to any of claims 1-6.
CN202110772313.9A 2021-07-08 2021-07-08 Block chain technology-based method for managing deposit duration bonds Withdrawn CN113450222A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110772313.9A CN113450222A (en) 2021-07-08 2021-07-08 Block chain technology-based method for managing deposit duration bonds

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110772313.9A CN113450222A (en) 2021-07-08 2021-07-08 Block chain technology-based method for managing deposit duration bonds

Publications (1)

Publication Number Publication Date
CN113450222A true CN113450222A (en) 2021-09-28

Family

ID=77815371

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110772313.9A Withdrawn CN113450222A (en) 2021-07-08 2021-07-08 Block chain technology-based method for managing deposit duration bonds

Country Status (1)

Country Link
CN (1) CN113450222A (en)

Similar Documents

Publication Publication Date Title
CN113822494B (en) Risk prediction method, device, equipment and storage medium
US20210264272A1 (en) Training method and system of neural network model and prediction method and system
CN110751286A (en) Training method and training system of neural network model
CN112380870A (en) User intention analysis method and device, electronic equipment and computer storage medium
CN114186132B (en) Information recommendation method and device, electronic equipment and storage medium
CN114077841A (en) Semantic extraction method and device based on artificial intelligence, electronic equipment and medium
CN112559687A (en) Question identification and query method and device, electronic equipment and storage medium
CN113706291A (en) Fraud risk prediction method, device, equipment and storage medium
Ferrando et al. Physics-Informed Machine Learning for Electricity Markets: A NYISO Case Study
CN113628043A (en) Complaint validity judgment method, device, equipment and medium based on data classification
CN117350794A (en) Day-ahead electricity price prediction method, device and medium based on multi-model fusion
CN117611133A (en) Large model reasoning acceleration method and system based on electric power operation detection service
CN110213239B (en) Suspicious transaction message generation method and device and server
Yuanfang Optimization of financial market risk prediction system based on computer data simulation and markov chain monte carlo
CN113450222A (en) Block chain technology-based method for managing deposit duration bonds
CN113450229A (en) Block chain technology-based method for managing deposit duration bonds
Lai et al. Chinese stock market prediction based on multifeature fusion and TextCNN
Xu et al. [Retracted] The Dissemination and Evaluation of Campus Ideological and Political Public Opinion Based on Internet of Things Monitoring
CN115271686B (en) Intelligent checking method and device for government affair data
CN117992598B (en) Demand response method, device, medium and equipment based on large model
CN114611712B (en) Prediction method based on heterogeneous federated learning, model generation method and device
CN113297849B (en) Financial pre-charge management method based on block chain technology
Duong et al. Ensemble Learning Technique with A Novelty Multi‑Source Information for Stock Price Movements
Zhang Sustainable Scientific and Technological Talents Recommendation Method Based on Recommendation Algorithm
Zhao et al. NASDAQ 100 Index Prediction Using LSTM And Sentiment Analysis

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
WW01 Invention patent application withdrawn after publication
WW01 Invention patent application withdrawn after publication

Application publication date: 20210928