Trading Method, Device and System for Identifying Broker Quotes Based on Artificial Intelligence

Abstract

The present invention discloses a trading method, device and system for identifying brokers' quotes based on artificial intelligence, which relates to the technical field of artificial intelligence. The method includes: obtaining the offline transaction details information sent by a currency broker through an associated instant messaging tool; calling a preset transaction data recognition model to perform transaction element recognition and format conversion processing on the offline transaction details information to obtain a standardized transaction intention order to be transacted, where the standardized transaction intention order is structured data in a preset data format; for the standardized transaction intention order, according to different quotation methods, converting it into a quotation order to be matched and sending it to a designated online trading platform, or converting it into a proposed price-responding quotation order to be matched and storing it in the aforementioned platform. The present invention effectively improves the applicable scope of intelligent trading, covers the transaction processing requirements of transaction personnel of currency brokers, and improves the trading efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of artificial intelligence, and particularly relates to a trading method, device and system for identifying broker quotes based on artificial intelligence. Background Art

[0002] With the trend of networking and globalization in the financial industry, in response to the rapid changes in the market and the business development of banks, a trading management system for inter-bank market transaction data management has emerged, such as the ComStar system. The ComStar system covers all business varieties of the domestic and foreign currency trading platforms of the China Foreign Exchange Trade System, the spot and repurchase transactions of the Shanghai and Shenzhen Stock Exchanges, the spot and derivative transactions of the Gold Exchange, the transfer and repurchase transactions of the Bill Exchange, the Treasury bond futures business of the China Financial Futures Exchange, as well as various offline businesses, achieving the integration of domestic and foreign currencies and the straight-through processing of the front, middle and back offices, greatly reducing manual operations and reducing operational risks. At the same time, the ComStar system can be seamlessly integrated with the trading platform and post-trading platform of the Foreign Exchange Trade System to achieve straight-through processing of trading strategies, pre-approval, real-time quota control, transaction confirmation and fund settlement.

[0003] The interbank market consists of the interbank lending market, bill market, bond market, foreign exchange market, gold market, etc. Money brokers (usually money brokerage companies) are important intermediary institutions in the interbank market, assuming the function of "information intermediary". Their main task is to provide correct and prompt trading information to facilitate the conclusion of transactions. Taking bond trading as an example, each money broker enters the bond trading information of the financial institutions it is responsible for into a platform (such as the platform of financial information service providers), displays it to the whole market, and then interested traders can contact the institutions through this information to fulfill the function of facilitating transactions. The broker quotes have greatly improved the trading efficiency of the market and solved the problem of information asymmetry, and have gradually become the mainstream bond transaction method in the interbank bond market. The intermediary quotes of traditional brokers are mainly released through specialized instant messaging tools / platforms, and the quote information is mainly displayed in text form. The information presented can include many elements such as bond varieties, codes, maturities, trading directions, trading prices, trading amounts, etc.; at the same time, the quote information is updated in real time, and new quote information is continuously introduced by refreshing the screen; the quotes of different trading brokers can be independently displayed in different windows, and different broker quotes can be viewed by switching windows. At present, for broker quotes, traders need to spend a long time to view and understand the trading elements, which affects the trading efficiency and is easy to miss trading opportunities. At the same time, due to the large number of bond varieties and maturities involved in trading quotes, the large number of trading institutions, the rapid change of quote information, and the complexity of quote message entries, traders must constantly search for the information they are concerned about in a large amount of quote texts, manually compare the scattered quote information, and also need to manually switch different windows. The efficiency is low and it is easy to miss quotes and the best prices, which is not conducive to quickly and accurately making trading value judgments and difficult to meet the trading requirements with strong competitiveness and high timeliness. Moreover, after traders and trading counterparts reach a trading intention in the later stage, they still need to manually enter the trading quote form in the trading system of the foreign exchange trading center to finally conclude the transaction, consuming a lot of labor costs and time costs.

[0004] In the context of the continuous advancement of the digital economy, in the face of the diverse changes in business models and the rising labor costs, major financial institutions have started to vigorously invest in the research and application of artificial intelligence (i.e., AI) in fields such as trading, risk control, and customer service. By using artificial intelligence technology to perform intelligent trading data processing, daily work can be made more efficient and accurate, reducing manual consumption and greatly meeting the refined business demands of financial institutions.

[0005] For example, Chinese Patent Application CN 202211158733.9 provides a list generation method, including: obtaining a natural language inquiry text sent by the counterparty's client; extracting transaction elements from the natural language inquiry text, where the transaction elements include: transaction direction, transaction term, and transaction amount; if an intention quotation form is determined to be generated based on the transaction elements, then generate the intention quotation form and send the intention quotation form to the counterparty's client; if the intention quotation form confirmation information sent by the counterparty's client is received, then send the transaction elements to the transaction system server so that the transaction system server generates a transaction quotation form. The above solution extracts the transaction elements of the inquiry through natural language processing, preliminarily determines whether to agree to the current inquiry and generates an intention quotation form based on the extracted transaction elements, and sends the transaction elements to the transaction system server so that the transaction system server generates a transaction quotation form according to the transaction elements, without manual entry of the transaction quotation form by the trader on this side, reducing the time consumed for finally reaching a transaction and reducing the consumption of labor costs. Another example is that Chinese Patent Application CN202310188187.1 provides a method for intelligent trading of AI trader inquiries, including the following steps: S1: The AI trader receives the other party's inquiry message and obtains the other party's identity; S2: The AI trader initiates multiple rounds of inquiries to the other party to obtain necessary transaction elements; S3: The AI trader transmits the necessary transaction elements to the foreign exchange market maker system, and the foreign exchange market maker system makes a quotation and feedbacks to the AI trader and the other party; S4: If the other party cancels the transaction, the process ends; if the other party confirms the quotation, then search for and confirm the counterparty; S5: The AI trader initiates multiple rounds of inquiries to the other party again to obtain the remaining non-necessary transaction elements; S6: The AI trader initiates a transaction according to the transaction method, obtains the information corresponding to the transaction method, and matches the information corresponding to the transaction method with all the information obtained by the AI trader. If the match is successful, the transaction is completed; if the match is not successful, manual intervention is transferred. The above solution can automatically screen trading intentions and perform trading matching by setting an AI trader (AI robot) between the inquirer and the foreign exchange market maker system, liberating human resources and improving the accuracy and timeliness of foreign exchange inquiry quotations. In addition, the above solution uses pre-trade credit risk control and operational risk control to enhance risk management capabilities: automatically query and pre-occupy the counterparty's credit limit before the transaction for pre-trade credit risk control, and automatically check the trading intention and the transaction elements of the actual order during the transaction to assist human traders in completing transaction confirmation and realizing operational risk control.

[0006] However, the above two solutions are only applicable to the application scenarios where online inquiry and quotation are directly carried out through the online client. For business transactions negotiated offline, after the transaction is reached, the trader still needs to manually enter the transaction quotation form in the trading system of the foreign exchange trading center to finally complete the transaction. Moreover, foreign exchange transactions may involve multiple parties, and different parties have personalized transaction processing requirements based on their own functions. It is difficult for the intelligent trading methods provided by the above solutions to cover the transaction processing requirements of the traders of currency brokers. On the other hand, the former does not set any risk control strategies, and although the latter considers risk management and control, it cannot conduct risk verification on the quotation information, which may lead to potential risks in the automated transactions reached. Summary of the Invention

[0007] The purpose of the present invention is to overcome the deficiencies of the prior art and provide a trading method, device and system for identifying broker quotations based on artificial intelligence. The present invention can perform automated transaction processing on the transaction data negotiated offline by currency brokers. By calling a preset transaction data recognition model, the transaction element recognition and format conversion processing are performed on the offline transaction details information of the transaction, and after obtaining the standardized transaction intention form to be concluded, according to different quotation methods, the foregoing standardized transaction intention form can be converted into a quotation form to be matched and then sent to a specified online trading platform, or converted into a quotation form to be counter-offered and then stored in the foregoing platform. No trader intervention is required in the above process, effectively improving the applicable scope of intelligent trading, covering the transaction processing requirements of the traders of currency brokers, and improving the trading efficiency. Further, risk verification is performed on the quotation information in the processing flow, reducing the potential risks of automated transactions.

[0008] To achieve the above objectives, the present invention provides the following technical solutions:

[0009] A trading method for identifying broker quotations based on artificial intelligence, including the steps of:

[0010] Obtain the offline transaction details information of a transaction sent by a currency broker through an associated instant messaging tool, and the offline transaction details information is unstructured data based on natural language;

[0011] Call a preset transaction data recognition model to perform transaction element recognition and format conversion processing on the offline transaction details information of the transaction, and obtain a standardized transaction intention form to be concluded, and the standardized transaction intention form is structured data in a preset data format;

[0012] For the foregoing standardized transaction intention form, according to different quotation methods, the foregoing standardized transaction intention form is converted into a quotation form to be matched and then sent to a specified online trading platform through a quotation processing module, or the foregoing standardized transaction intention form is converted into a quotation form to be counter-offered and then stored in the foregoing platform.

[0013] Further, the quotation methods include dialogue quotation and request-for-quotation methods. Corresponding to the dialogue quotation and request-for-quotation methods, different quotation processing modes are configured for the standardized trading intention form, and the quotation processing modes include a dialogue quotation mode and a request-for-quotation mode;

[0014] In the dialogue quotation mode, select to convert the foregoing standardized trading intention form into a formal quotation for sending or into a pending matching quotation form;

[0015] In the request-for-quotation mode, select to convert the foregoing standardized trading intention form into a formal quotation for sending or into a pending matching quotation form for proposed counter-offer.

[0016] Further, in the dialogue quotation mode, the quotation processing module is configured as follows:

[0017] When the party itself is the initiator of the dialogue quotation, the party itself sends the quotation of this transaction to the counterparty; determine whether the counterparty has performed a confirmation or rejection operation within a preset first time threshold range; when the counterparty confirms within the foregoing first time threshold range, determine that this transaction is concluded, convert the foregoing standardized trading intention form into a formal quotation for sending, so as to conclude a formal transaction on the specified domestic currency trading platform; when the counterparty does not perform a confirmation or rejection operation within the foregoing first time threshold range, obtain the closing time of the specified domestic currency trading platform, and when it is monitored that the closing time of the specified domestic currency trading platform arrives, prompt the party itself that the quotation of this transaction has expired; when the counterparty rejects within the foregoing first time threshold range, determine that the quotation of this transaction is rejected, and update the status of this transaction to rejected;

[0018] When the party itself is the recipient of the dialogue quotation, convert the foregoing standardized trading intention form into a pending matching quotation form; when the counterparty initiates a quotation is collected, automatically match the received quotation information with the transaction elements of the foregoing pending matching quotation form, and when the elements match, determine that this transaction is concluded, and convert the foregoing pending matching quotation form into a formal quotation form, so as to conclude a formal transaction on the specified domestic currency trading platform.

[0019] Further, in the request-for-quotation mode, the quotation processing module is configured as follows:

[0020] When the party itself is the initiator of the request-for-quotation, convert this transaction task into a request-for-quotation for sending, and after the counterparty clicks to confirm, this transaction is concluded;

[0021] When the party itself is the recipient of the request for quotation, convert the foregoing standardized transaction intention form into a quotation form to be counter-offered and awaiting matching; after receiving the request for quotation information from the counterparty, automatically match the received request for quotation information with the transaction elements in the foregoing quotation form to be counter-offered and awaiting matching. When the elements are matched, determine that the transaction is successfully matched and make a counter-offer to the counterparty.

[0022] For the foregoing counter-offer, when the counterparty performs a confirmation operation within the preset second time threshold range, determine that the transaction is concluded. At this time, convert the foregoing quotation form to be counter-offered and awaiting matching into an official quotation form, so as to conclude an official transaction on the designated domestic currency trading platform; when the counterparty does not perform a confirmation operation within the foregoing second time threshold range, prompt the party itself that the quotation for this transaction has expired.

[0023] Furthermore, it further includes a risk verification module, which is used to verify the quotation information in the stage of identifying transaction elements, generating a standardized transaction intention form, and / or the quotation processing stage to determine whether there is a risk in the quotation information; the verification items configured in the risk verification module at least include a price deviation degree verification item and a price deviation value verification item.

[0024] The risk verification module is configured to: obtain the quotation price in the quotation information in the above stage, calculate the price deviation degree and the price deviation value between the foregoing quotation price and the preset reference price. When any one of the price deviation degree and the price deviation value exceeds the preset limit value, determine that there is a risk in the quotation information and terminate the transaction process; when both the price deviation degree and the price deviation value are less than or equal to the preset limit value, determine that there is no risk in the quotation information and perform subsequent processing.

[0025] Furthermore, for the foregoing transaction data recognition model, configure a list of money brokers who can apply the transaction data recognition model; at this time, before calling the foregoing transaction data recognition model to identify the offline transaction details information, it further includes the steps:

[0026] Obtain the identity ID information of the money broker who sends the offline transaction details information, and determine whether the money broker is in the foregoing broker list according to the foregoing identity ID information. When it is determined that the money broker is in the broker list, call the foregoing transaction data recognition model.

[0027] Furthermore, the transaction data recognition model includes a natural language parsing unit and a format conversion unit; a format specification setting unit is configured for the preset data format corresponding to the standardized transaction intention form, and the transaction element requirements and data structure requirements of the standardized transaction intention form are set through the format specification setting unit.

[0028] The natural language parsing unit is used to perform natural language parsing on the offline transaction details information to identify all transaction element information of this transaction.

[0029] The format conversion unit is configured to extract multiple pieces of transaction element information that match the requirements of the transaction elements from all the identified transaction elements according to the requirements of the transaction elements in the format specification of the foregoing standardized transaction intention form; and then generate a standardized transaction intention form with elements and structure conforming to the foregoing format specification based on the extracted transaction element information and the data structure requirements in the foregoing format specification.

[0030] Furthermore, the associated instant messaging tool is the iDeal dialogue tool provided by CFETS; the designated online trading platform is the domestic currency trading platform of CFETS.

[0031] When the trading object is a bond, the transaction elements required to be configured in the standardized transaction intention form include the compensation period of this transaction, bond code, clearing speed, our party's direction, yield rate, trading volume, counterparty institution, counterparty trader, dialogue institution, dialogue trader, and order placement method.

[0032] The present invention also provides a trading automatic processing device based on artificial intelligence to identify brokers' quotes, and the device includes the following structures:

[0033] An information collection module, configured to obtain the off-line transaction details information of a transaction sent by a money broker through the associated instant messaging tool, and the off-line transaction details information is unstructured data based on natural language;

[0034] An intention generation module, configured to call a preset transaction data recognition model to perform transaction element recognition and format conversion processing on the off-line transaction details information of this transaction, and obtain a standardized transaction intention form to be transacted, and the standardized transaction intention form is structured data using a preset data format;

[0035] A quote processing module, configured to convert the foregoing standardized transaction intention form into a quote to be matched and send it to the designated online trading platform according to different quote methods for the foregoing standardized transaction intention form, or convert the foregoing standardized transaction intention form into a quote to be counter-offered and stored in the designated online trading platform.

[0036] The present invention also provides a trading execution system, and the trading execution system can directly access the domestic currency trading platform of CFETS; wherein, the foregoing trading automatic processing device is provided in the trading execution system.

[0037] Due to the adoption of the above technical solutions, compared with the prior art, the present invention has the following advantages and positive effects by way of example: The present invention can perform automated processing on the transaction data negotiated offline by currency brokers. By calling a preset transaction data recognition model, it performs transaction element recognition and format conversion processing on the offline transaction details information of the transaction. After obtaining the standardized transaction intention order to be concluded, according to different quotation methods, the foregoing standardized transaction intention order can be converted into a quotation order to be matched and sent to a designated online trading platform, or converted into a quotation order to be counter-offered and stored in the foregoing platform. No trader is involved in the above process, effectively improving the applicable scope of intelligent trading, covering the transaction processing requirements of the traders of currency brokers, and improving the trading efficiency.

[0038] Furthermore, risk verification is performed on the quotation information in the processing flow, reducing the potential risks of automated trading. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a flowchart of a trading method for identifying brokers' quotations based on artificial intelligence provided by an embodiment of the present invention.

[0040] Figure 2 It is an information transmission diagram of the trading method provided by an embodiment of the present invention.

[0041] Figure 3 It is a module structure diagram of a trading automatic processing device provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0042] The following further details the trading method, device, and system for identifying brokers' quotations based on artificial intelligence disclosed by the present invention in conjunction with the accompanying drawings and specific embodiments. It should be noted that well-known technologies (including methods and devices) in the relevant fields may not be discussed in detail, but in appropriate cases, the above well-known technologies are regarded as part of the specification. At the same time, other examples of the exemplary embodiments may have different values. The structures, ratios, sizes, etc. shown in the drawings of this specification are only used to cooperate with the content disclosed in the specification for those familiar with this technology to understand and read, and are not used to limit the limiting conditions for the implementation of the invention.

[0043] In the description of the embodiments of the present application, " / " means "or", and "and / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" means: A exists alone, B exists alone, and A and B exist simultaneously. In the description of the embodiments of the present application, "a plurality of" means two or more.

[0044] Term Explanation:

[0045] 1) CFETS: The full name is China Foreign Exchange Trade System, which is the main trading system in the Chinese foreign exchange market. The domestic currency trading platform (system) provided by CFETS covers various trading varieties such as multiple types of interest rates and credits, and supports multiple trading methods such as inquiry, market-making quotes, request-for-quote, and bilateral credit matching.

[0046] 2) iDeal Dialogue Tool: The iDeal Dialogue Tool is an instant messaging tool (or instant messaging platform) launched by the China Foreign Exchange Trade System CFETS for the inter-bank domestic and foreign currency markets. It is an Internet product integrating a series of business and technical features such as real-name authentication, compliance supervision, trading assistance, system integration, and open ecosystem; at the same time, it also provides data information services such as market data, benchmark data, and iData data mining products.

[0047] 3) ComStar System is a trading management system for trading data management in the inter-bank market. The ComStar System consists of multiple functional modules. Through the combination of different functional modules, it supports different business lines and builds a management system for banks from information acquisition, front-office trading analysis and management, middle-office risk control to back-office straight-through processing of transactions.

[0048] 4) BERT-CRF Model: The BERT-CRF model is a sequence labeling model that combines the pre-trained language model BERT (Bidirectional Encoder Representations from Transformers) and the Conditional Random Field, and is commonly used in natural language processing. BERT is a pre-trained model based on Transformer, which can effectively extract and process semantic information in text; CRF is a model based on conditional random field, which can consider the orderliness of features in text to achieve accurate identification of new entities. The BERT-CRF model is essentially a CRF model, but it uses the BERT model to train the emission matrix in the CRF model. It extracts features from text data through the BERT model, and then inputs the extracted features into the CRF model for the identification and classification of new entities.

[0049] 5) SimCSE: The full name is Simple Contrastive Learning of Sentence Embeddings. The SimCSE model is one of the commonly used models for obtaining text vector representations. By integrating contrastive learning, it can better obtain the expression of text vectors and can be used to improve the performance of language models in sentence representation tasks. Its basic principle is contrastive learning.

[0050] The technical concept and solution of the present invention will be introduced below according to exemplary application scenarios. Embodiment

[0051] Currently, the prior art has provided the use of artificial intelligence (AI) robots in trading systems for intelligent inquiry, automatic response, and intention achievement. Among them, the AI robot uses natural language processing (NLP) and applies a machine learning (ML) model to automatically reply to the inquiries of trading counterparts, actively inquire about trading counterparts, automatically reach intention orders, and automatically generate corresponding quotation sheets (the AI robot can automatically generate corresponding formal trading quotation sheets in the foreign exchange trading center system). However, the above prior art is limited to the application scenario of directly conducting online inquiry and quotation through robots, and cannot automatically generate intention orders and quotation sheets for business transactions negotiated offline. At the same time, the above solution is difficult to cover the trading processing requirements of the trading personnel of currency brokers.

[0052] The present invention makes improvements to the above prior art, configures a deal data recognition model for identifying the deal data of brokers, automatically identifies the deal data of brokers, and can directly access the local currency trading platform of CFETS to conclude a formal transaction.

[0053] Specifically, refer to Figure 1 As shown, a trading method for identifying broker quotations based on artificial intelligence provided in this embodiment includes the following steps.

[0054] S100, obtain the offline deal details information of a transaction sent by a currency broker through an associated instant messaging tool, where the offline deal details information is unstructured data based on natural language.

[0055] Preferably, the associated instant messaging tool is the iDeal dialogue tool provided by CFETS. In this step, the currency broker can send the deal details information negotiated offline (intention reached offline) to the transaction automatic processing device through the iDeal dialogue tool, and a broker deal data recognition model is configured in the transaction automatic processing device to identify the transaction elements and convert the data format of the offline deal obvious information.

[0056] The offline transaction details information is unstructured data based on natural language. For example, it can be the chat conversation records between currency broker traders and financial institution traders. An offline transaction data collection interface can be set on the iDeal dialogue tool for brokers to input the text content of the offline negotiated transaction data; or for brokers to import the offline negotiated transaction data file, and the imported transaction data file can be a text file or an audio file. When the imported transaction data file is an audio file, the audio file is first converted into text to obtain the corresponding transaction data text content, and then the broker transaction data recognition model is used to identify transaction elements and convert the data format of the transaction data text content.

[0057] Optionally, the offline broker transaction data sent through the iDeal dialogue tool can be stored in the associated big data warehouse to support retrospective tracing and compliance review, thus meeting regulatory requirements.

[0058] S200, call the preset transaction data recognition model to identify transaction elements and convert the format of the offline transaction details information of this transaction, and obtain a standardized transaction intention order to be transacted. The standardized transaction intention order is structured data in a preset data format.

[0059] S300, for the aforementioned standardized transaction intention order, according to different quotation methods, convert the aforementioned standardized transaction intention order into a quotation order to be matched through the quotation processing module and send it to the specified online trading platform; or convert the aforementioned standardized transaction intention order into a counter-offer quotation order to be matched and store it in the aforementioned platform.

[0060] The specified online trading platform can be the domestic currency trading platform of CFETS.

[0061] In this embodiment, for the aforementioned transaction data recognition model, a list of currency brokers to which the transaction data recognition model can be applied can also be pre-configured. Specifically, the transaction automatic processing device of this embodiment can include an initialization unit, and the list of currency brokers is configured through the initialization unit. The list of currency brokers preferably takes the form of a table, that is, a currency broker list is formed.

[0062] Based on the list of currency brokers, users can preset the list of brokers that can receive the application of the transaction data recognition model as needed. Each broker has its own unique identity ID in the list of currency brokers. If a broker is in the list of currency brokers, the broker can use the transaction data recognition model to identify and convert offline transaction data; if a broker is not in the list of currency brokers, the broker cannot use the transaction data recognition model.

[0063] At this time, before step S200, the method may further include the steps of: obtaining the identity ID information (i.e., identity identification information, where one broker corresponds to a unique identity ID) of the money broker that sends the off-line transaction details information, determining whether the money broker is in the aforementioned broker list according to the aforementioned identity ID information, and when it is determined that the money broker is in the broker list, calling the aforementioned transaction data recognition model. When it is determined that the broker is not in the broker list, a reminder that the money broker does not have the authority to recognize transaction data may be output, and the model calling process may be ended.

[0064] In this embodiment, the transaction automatic processing device may further include a format specification setting unit corresponding to the preset data format configuration of the standardized transaction intention form. Through the format specification setting unit, the user may set the formatting requirements of the standardized transaction intention form, including transaction element requirements (data content requirements) and data structure requirements.

[0065] Specifically, the transaction data recognition model may include a natural language parsing unit and a format conversion unit.

[0066] The natural language parsing unit is configured to perform natural language parsing on the off-line transaction details information to identify all transaction element information of this transaction.

[0067] Preferably, when identifying the transaction elements in the broker's transaction details information, the BERT-CRF model may be used to extract and identify the transaction elements. The latest conversation terms of the broker's transaction data are trained by BERT-CRF to be close to the real transaction business scenario. Using the trained BERT-CRF model to identify the transaction element information in the natural language text can improve the accuracy of extracting elements. The BERT-CRF model can exclude interference factors according to the context semantic information. The specific BERT-CRF model training process and entity (transaction element) identification method refer to the prior art and will not be elaborated here.

[0068] Optionally, in order to further improve the accuracy of transaction element recognition, on the basis of using the BERT-CRF model for entity recognition, a method combining string matching and the SimCSE model may also be used to perform institutional entity mapping, mapping the entities (transaction elements) included in the off-line negotiated transaction data text to the corresponding entities (transaction elements) in the institutional database. The same transaction element may have different names in the off-line transaction data - for example, in the transaction details, there may be Chinese names, Chinese abbreviations, English names, English abbreviations, English capitals, English lower cases, etc. of the same transaction element. By combining the string matching and the contrast learning function of the SimCSE model, the relevant names of the same transaction element in the transaction data can be obtained more accurately, and then mapped to the corresponding transaction elements in the institutional database.

[0069] The format conversion unit is configured to extract multiple pieces of transaction element information that match the requirements of the transaction elements from all the identified transaction elements according to the requirements of the transaction elements in the format specification of the foregoing standardized transaction intention form; then, based on the extracted transaction element information and the data structure requirements in the foregoing format specification, generate a standardized transaction intention form whose elements and structure conform to the foregoing format specification.

[0070] Taking the bond as the trading object as an example, at this time, the transaction elements required to be configured in the standardized transaction intention form may include the compensation period, bond code, clearing speed, our side direction, yield, trading volume, counterparty institution, counterparty trader, dialogue institution, dialogue trader, and order sending method of this transaction, etc. According to these transaction elements, after extracting the transaction data content corresponding to the above transaction elements from the identified broker transaction details information - by way of example and not limitation, for example, for a certain bond transaction, the transaction data content corresponding to the transaction element "compensation period" is "3 years", the transaction data content corresponding to the transaction element "bond code" is "010**3" (the unique number used to identify and trade the bond), the transaction data content corresponding to the transaction element "clearing speed" is "T+0", etc., and then according to the data structure requirements in the format specification, generate a standardized transaction intention form. The standardized transaction intention form uses the foregoing preset data format to record the above transaction element field names and the transaction data content corresponding to each transaction element.

[0071] In step S300, according to different quotation methods, the quotation processing module can choose to directly send the standardized transaction intention form converted into a quotation to be matched to a specified online trading platform - such as the domestic currency trading platform of CFETS, or choose to convert the foregoing standardized transaction intention form into a quotation to be matched with a proposed price and store it in the foregoing domestic currency trading platform.

[0072] Specifically, as shown in Figure 2 The quotation methods include the dialogue quotation and the request-for-quotation method. Corresponding to the dialogue quotation and the request-for-quotation method, different quotation processing modes are configured for the standardized transaction intention form, and the quotation processing modes include the dialogue quotation mode and the request-for-quotation mode.

[0073] In the dialogue quotation mode, choose to convert the foregoing standardized transaction intention form into a formal quotation for sending or into a quotation to be matched.

[0074] In the request-for-quotation mode, choose to convert the foregoing standardized transaction intention form into a formal quotation for sending or into a quotation to be matched with a proposed price.

[0075] Specifically, in the said dialogue quotation mode, the quotation processing module is configured as follows: when the home party is the initiator of the dialogue quotation, the home party sends the quotation for this transaction to the counterparty; determine whether the counterparty has performed a confirmation or rejection operation within a preset first time threshold range; when the counterparty confirms within the aforesaid first time threshold range, determine that this transaction is concluded, convert the aforesaid standardized transaction intention form into a formal quotation and send it to the designated domestic currency trading platform, so as to conclude a formal transaction on the designated domestic currency trading platform; when the counterparty does not perform a confirmation or rejection operation within the aforesaid first time threshold range, obtain the closing time of the designated domestic currency trading platform, and when it is monitored that the closing time of the designated domestic currency trading platform arrives, prompt the home party that the quotation for this transaction has expired; when the counterparty rejects within the aforesaid first time threshold range, determine that the quotation for this transaction is rejected, and update the status of this transaction to rejected.

[0076] When the home party is the recipient of the dialogue quotation, convert the aforesaid standardized transaction intention form into a quotation to be matched (transaction task to be matched) and store it in the platform's task list to be matched; when it is collected that the counterparty initiates a quotation, automatically match the received quotation information with the transaction elements of the aforesaid quotation to be matched. When the elements are matched, determine that this transaction is concluded, and convert the aforesaid stored quotation to be matched into a formal quotation form, so as to conclude a formal transaction on the designated domestic currency trading platform.

[0077] In this embodiment, the matching elements may include transaction counterparty institution, counterparty, trading volume, trading date, clearing speed, yield rate and other transaction elements.

[0078] In the said request for quotation mode, the quotation processing module is configured as follows: when the home party is the initiator of the request for quotation, convert this transaction task into a request for quotation and send it. After the counterparty clicks to confirm, this transaction is concluded. When the home party is the recipient of the request for quotation, convert the aforesaid standardized transaction intention form into a quotation to be matched for a proposed price reply and store it in the platform; after receiving the request for quotation information from the counterparty, automatically match the received request for quotation information with the transaction elements in the aforesaid quotation to be matched for a proposed price reply. When the elements are matched, determine that the transaction is successfully matched and reply with a price to the counterparty. The matching elements may include transaction counterparty institution, counterparty, trading volume, trading date, clearing speed, yield rate, etc.

[0079] For the aforementioned counter-offer, when the counterparty performs a confirmation operation within the preset second time threshold range, the transaction is determined to be completed. At this time, the aforementioned proposed counter-offer to be matched quotation sheet is converted into a formal quotation sheet and sent, so as to complete the formal transaction on the designated local currency trading platform. When the counterparty does not perform a confirmation operation within the aforementioned second time threshold range, it will remind this party that the quotation for the transaction has expired.

[0080] As an example but not limitation, for example, the first time threshold and the second time threshold can be configured as 2 hours. It should be noted that the specific time lengths of the first time threshold and the second time threshold can be configured according to actual transaction needs, and the two can be configured to be the same or different, which should not be used as a limitation to the present invention.

[0081] In another implementation of this embodiment, a risk check module may also be included. The risk check module is used to check the quotation information in the stage of identifying transaction elements, generating standardized transaction intention forms and / or quotation processing to determine whether the quotation information has risks. The check items configured in the risk check module include at least a price deviation check item and a price deviation value check item. When the quotation exceeds the limit, the transaction process will be forcibly terminated.

[0082] Specifically, the risk check module is configured to: obtain the quoted price in the quotation information in the above stage, calculate the price deviation and price deviation value between the above quoted price and the preset reference price, and when any one of the price deviation and the price deviation value exceeds the preset limit, determine that the quotation information is risky and terminate the transaction process; when the price deviation and the price deviation value are both less than or equal to the preset limit, determine that the quotation information is risk-free and execute subsequent processing.

[0083] In one implementation of this embodiment, for the quotation to be matched, an approval step is also configured, that is, the quotation to be matched is sent to CFETS after approval; for the quotation to be matched with a proposed counter-offer, an approval step can also be configured, that is, the quotation to be matched with a proposed counter-offer is sent to the platform for storage after approval. The approval step can be automatic approval by the system (the relevant approval model is configured in the system) or manual approval or a combination of the two, which is not limited here.

[0084] Preferably, for quotation information determined by the risk verification module to be risk-free (within the risk control limit), the generated quotation sheet can be directly uploaded to the local currency trading platform of CFETS after prior approval according to the requirements of the institution.

[0085] See also Figure 3 As shown, another embodiment of the present invention also provides an automatic transaction processing device based on artificial intelligence to identify broker quotations.

[0086] The transaction automatic processing device includes an information collection module, an intention generation module, and a quotation processing module.

[0087] The information collection module is used to obtain the off-line transaction details information of a transaction sent by a money broker through an associated instant messaging tool, and the off-line transaction details information is unstructured data based on natural language.

[0088] The intention generation module is used to call a preset transaction data recognition model to perform transaction element recognition and format conversion processing on the off-line transaction details information of this transaction, and obtain a standardized transaction intention form to be transacted. The standardized transaction intention form is structured data in a preset data format.

[0089] The quotation processing module is used to, for the aforementioned standardized transaction intention form, according to different quotation methods, convert the aforementioned standardized transaction intention form into a quotation form to be matched and send it to a specified online trading platform, or convert the aforementioned standardized transaction intention form into a quotation form to be matched with a proposed price return and store it in the specified online trading platform.

[0090] Specifically, the quotation methods may include the dialogue quotation and the request quotation method. Corresponding to the dialogue quotation and the request quotation method, different quotation processing modes are configured for the standardized transaction intention form, and the quotation processing modes include the dialogue quotation mode and the request quotation mode.

[0091] In the dialogue quotation mode, select to convert the aforementioned standardized transaction intention form into a formal quotation for sending or into a quotation form to be matched.

[0092] In the request quotation mode, select to convert the aforementioned standardized transaction intention form into a formal quotation for sending or into a quotation form to be matched with a proposed price return.

[0093] For other technical features, refer to the description of the previous embodiments, which will not be elaborated here.

[0094] Another embodiment of the present invention further provides a transaction execution system. The transaction execution system can be set in the ComStar system, and the transaction execution system can directly access the local currency trading platform of CFETS.

[0095] A transaction automatic processing device is set in the transaction execution system. The transaction automatic processing device includes an information collection module, an intention generation module, and a quotation processing module.

[0096] The information collection module is used to obtain the off-line transaction details information of a transaction sent by a money broker through an associated instant messaging tool, such as through the iDeal dialogue tool. The off-line transaction details information is unstructured data based on natural language.

[0097] The intention generation module is used to call a preset transaction completion data recognition model to perform transaction element recognition and format conversion processing on the offline transaction completion details of this transaction, so as to obtain a standardized transaction intention form to be completed. The standardized transaction intention form is structured data in a preset data format.

[0098] The quotation processing module is used to convert the aforesaid standardized transaction intention form into a quotation form to be matched and send it to a designated online trading platform according to different quotation methods for the aforesaid standardized transaction intention form, or convert the aforesaid standardized transaction intention form into a counter-offer quotation form to be matched and store it in the platform.

[0099] In this way, through the ComStar system, it is possible to automatically identify the offline negotiated transaction completion data sent by the broker through the iDeal dialogue tool, obtain the standardized transaction intention to be completed, and then perform quotation sending processing according to different quotation methods and the feedback information of the counterparty. After the transaction is completed, it can be directly uplinked to the CFETS RMB trading platform to conclude a formal transaction without the intervention of a trader, improving the trading efficiency.

[0100] For other technical features, reference can be made to the description of the foregoing embodiments and will not be elaborated herein.

[0101] The present invention also provides a computer-readable storage medium for storing a computer program executable by a processing unit. When the computer program is executed by the processing unit, the method described above is implemented.

[0102] The storage medium may include various media capable of storing program codes, such as a USB flash drive, a mobile hard disk, a read only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disc.

[0103] For other technical features, reference can be made to the description of the foregoing embodiments and will not be elaborated herein.

[0104] In the foregoing description, the disclosure of the present invention is not intended to limit itself to these aspects. Rather, within the scope of the object of the present disclosure, the components can be selectively and operably combined in any number. Additionally, terms such as "comprising", "including", and "having" should be construed as inclusive or open by default, rather than exclusive or closed, unless they are explicitly defined to the contrary. All technical, scientific, or other terms conform to the meanings understood by those skilled in the art, unless they are defined to the contrary. Common terms found in dictionaries should not be construed too idealistically or too unrealistically in the context of the relevant technical documents, unless the present disclosure explicitly defines them as such. Any changes or modifications made by those of ordinary skill in the art to the present invention based on the foregoing disclosure fall within the scope of protection of the claims.

Claims

1. A trading method based on artificial intelligence for identifying brokers' quotes, characterized in that Including the steps: Obtain the off-line transaction details information of a transaction sent by a currency broker through an instant messaging tool associated with a specified online trading platform, which is unstructured data based on natural language; an off-line transaction data collection interface is set on the instant messaging tool for the broker to input the text content of the off-line negotiated transaction data or import the off-line negotiated transaction data file; Call a preset transaction data recognition model to perform transaction element recognition and format conversion processing on the foregoing off-line transaction details information, and obtain a to-be-completed standardized transaction intention form, which is structured data in a preset data format; According to different quotation methods, the foregoing standardized transaction intention form is converted into a to-be-matched quotation form by a quotation processing module and then sent to the foregoing platform, or the foregoing standardized transaction intention form is converted into a to-be-counter-offered to-be-matched quotation form and then stored in the foregoing platform; wherein, the quotation methods include dialogue quotation and request quotation, corresponding to a dialogue quotation mode and a request quotation mode; In the dialogue quotation mode, when the party itself is the recipient of the dialogue quotation, convert the standardized transaction intention form into a to-be-matched quotation form and store it in the to-be-matched task list of the platform; when receiving a quotation initiated by the counterparty to the transaction, automatically match the received quotation information with the transaction elements of the to-be-matched quotation form, and when the elements are matched, determine that the transaction is concluded, and convert the to-be-matched quotation form into a formal quotation form; In the request quotation mode, when the party itself is the recipient of the request quotation, convert the standardized transaction intention form into a to-be-counter-offered to-be-matched quotation form; after receiving the request quotation information of the counterparty to the transaction, automatically match the received request quotation information with the transaction elements in the to-be-counter-offered to-be-matched quotation form, and when the elements are matched, determine that the transaction is successfully matched and make a counter-offer to the counterparty to the transaction.

2. The method according to claim 1, characterized in that, In the dialogue quotation mode, the quotation processing module is configured as follows: When the party itself is the initiator of the dialogue quotation, send the quotation of this transaction to the counterparty to the transaction; judge whether the counterparty to the transaction has performed a confirmation or rejection operation within a preset first time threshold range; when the counterparty to the transaction confirms within the foregoing first time threshold range, determine that the transaction is concluded, convert the foregoing standardized transaction intention form into a formal quotation and send it, so as to conclude a formal transaction on the specified domestic currency trading platform; when the counterparty to the transaction does not perform a confirmation or rejection operation within the foregoing first time threshold range, obtain the closing time of the specified domestic currency trading platform, and when it is monitored that the closing time of the specified domestic currency trading platform arrives, prompt the party itself that the quotation of this transaction has expired; when the counterparty to the transaction rejects within the foregoing first time threshold range, determine that the quotation of this transaction is rejected, and update the status of this transaction to rejected.

3. The method according to claim 1, characterized in that, In the request quotation mode, the quotation processing module is configured as follows: When the party itself is the initiator of the request quotation, convert this transaction task into a request quotation and send it, and when the counterparty to the transaction clicks to confirm, this transaction is concluded. When the party is the recipient of the request for quotation, for the aforesaid counter-offer, when the counterparty conducts a confirmation operation within the preset second time threshold range, it is determined that the transaction is concluded. At this time, the aforesaid counter-offer waiting-for-matching quotation form is converted into a formal quotation form, so as to conclude a formal transaction on the designated domestic currency trading platform; when the counterparty does not conduct a confirmation operation within the aforesaid second time threshold range, the party is prompted that the quotation for this transaction has expired.

4. The method according to claim 2 or 3, characterized in that: It further includes a risk verification module, which is used to verify the quotation information in the transaction element identification stage, the generation stage of the standardized transaction intention form and / or the quotation processing stage to judge whether there is a risk in the quotation information; the verification items configured in the risk verification module at least include a price deviation degree verification item and a price deviation value verification item; The risk verification module is configured to: obtain the quotation price in the quotation information in the above stages, calculate the price deviation degree and the price deviation value between the aforesaid quotation price and the preset reference price, and when any one of the price deviation degree and the price deviation value exceeds the preset limit value, it is determined that there is a risk in the quotation information and the transaction process is terminated; when both the price deviation degree and the price deviation value are less than or equal to the preset limit value, it is determined that there is no risk in the quotation information and subsequent processing is executed.

5. The method according to claim 1, wherein: For the aforesaid transaction data identification model, a list of money brokers capable of applying the transaction data identification model is configured; at this time, before calling the aforesaid transaction data identification model to identify the off-line transaction details information, it further includes the steps: Obtain the identity ID information of the money broker sending the off-line transaction details information, and judge whether the money broker is in the aforesaid broker list according to the aforesaid identity ID information. When it is determined that the money broker is in the broker list, call the aforesaid transaction data identification model.

6. The method according to claim 1, wherein: The transaction data identification model includes a natural language parsing unit and a format conversion unit; a format specification setting unit is configured corresponding to the preset data format of the standardized transaction intention form, and the transaction element requirements and data structure requirements of the standardized transaction intention form are set through the format specification setting unit; The natural language parsing unit is used to perform natural language parsing on the off-line transaction details information to identify all transaction element information of this transaction; The format conversion unit is used to extract multiple transaction element information that matches the transaction element requirements from the aforesaid identified all transaction elements according to the transaction element requirements in the format specification of the aforesaid standardized transaction intention form; then, based on the extracted transaction element information and the data structure requirements in the aforesaid format specification, generate a standardized transaction intention form whose elements and structure conform to the aforesaid format specification.

7. The method according to claim 6, characterized in that: The associated instant messaging tool is the iDeal dialogue tool provided by CFETS; the designated online trading platform is the domestic currency trading platform of CFETS; When the transaction object is a bond, the transaction elements required to be configured in the standardized transaction intention form include the compensation period of this transaction, bond code, clearing speed, the direction of the party, yield, trading volume, counterparty institution, counterparty trader, dialogue institution, dialogue trader and order placement method.

8. An automatic trading processing device for identifying broker quotes based on artificial intelligence according to the method described in claim 1, characterized in that Include: An information collection module, which is used to obtain the off-line transaction details information of a transaction sent by a money broker through an instant messaging tool associated with a specified online trading platform, and the information is unstructured data based on natural language; An intention generation module, which is used to call a preset transaction data recognition model to identify transaction elements and perform format conversion processing on the foregoing off-line transaction details information, so as to obtain a standardized transaction intention form to be transacted, and the form is structured data in a preset data format; A quotation processing module, which is used to convert the foregoing standardized transaction intention form into a quotation form to be matched and then send it to the foregoing platform according to different quotation methods, or convert the foregoing standardized transaction intention form into a quotation form to be counter-offered and then store it in the foregoing platform.

9. A trading execution system, the trading execution system can directly access the domestic currency trading platform of CFETS, and is characterized in that: The trading execution system is provided with the trading automatic processing device described in claim 8.

Images