An AI application call count counting method

Abstract

The application discloses a kind of AI application calling count statistics method, in AI application calling process, when different AI entities interact, log is recorded respectively, including token information based on asymmetric encryption technology;Different AI entities generate corresponding bill according to log, reconcile when billing dispute occurs, and the correct AI application calling times are counted, so as to correct billing;The AI entity includes AI application consumer, AI platform service provider, AI technology supplier.The application realizes the statistics of AI application calling count based on token and asymmetric encryption technology, records AI application calling bill by three parties, when billing dispute occurs, the correct AI application calling times statistics are obtained by comparing the bill of three parties, prevent the statistical error of AI application calling count, realize the accurate statistics of billing according to AI application calling.

Description

Technical Field

[0001] This invention belongs to the field of cloud computing and AI technology, and specifically relates to a statistical method for counting AI application calls. Background Technology

[0002] With the maturity of cloud computing and AI technologies, industrial digitalization is evolving in depth, and intelligent AI applications are gradually being implemented, accelerating the digital transformation of enterprises.

[0003] The digital ecosystem of the AI ​​industry is gradually becoming clearer, and three types of companies are emerging. The first type consists of AI application consumers, who purchase AI applications based on their own business needs, typically traditional industry companies. The second type consists of AI technology providers, who develop and deploy AI applications based on their own AI technologies for customers to purchase and use, typically internet companies and software vendors. The third type consists of AI platform service providers, who integrate AI applications from numerous AI technology providers. On the one hand, they provide platform clients (AI application consumers) with services such as querying, deployment, operation and maintenance, billing, and customer service; on the other hand, they help AI technology providers expand their business and share revenue with them, thus generating income for themselves.

[0004] There are three billing methods for AI applications: The first is that customers can buy the AI ​​application outright and use it anytime and in any way without having to pay again; the second is a customer subscription model, where customers pay a one-time fee within a time frame agreed upon by both parties and can use it as they please, with the option to renew or terminate the subscription service upon expiration; the third is to pay based on the number of times the AI ​​application is used, which can either price the application for each use or offer tiered pricing based on the number of uses, but both methods require tracking the number of uses.

[0005] The first and second billing methods are relatively simple and rarely cause billing disputes. However, the third method, billing by the number of calls, can lead to disputes due to discrepancies in the statistical data collected by the caller and the callee. The involvement of AI platform service providers makes billing by the number of calls even more complex and difficult to gain the full trust of all parties. In extreme cases, AI technology providers may be motivated to record more calls to increase revenue, while AI platform providers and AI application consumers may be motivated to record fewer calls to reduce costs. Summary of the Invention

[0006] The technical problem to be solved by this invention is to address the shortcomings of the prior art by providing a statistical method for counting AI application calls. This method is based on token and asymmetric encryption technology to count AI application calls. By having three parties record AI application call bills separately, in the event of a billing dispute, the correct AI application call count can be obtained by comparing the bills of the three parties, thus preventing errors in the counting of AI application calls and achieving accurate statistics on billing based on AI application calls.

[0007] To achieve the above-mentioned technical objectives, the technical solution adopted by the present invention is as follows:

[0008] A statistical method for counting AI application calls includes:

[0009] During the AI ​​application invocation process, logs are recorded separately when different AI entities interact, including token information based on asymmetric encryption technology;

[0010] Different AI entities generate corresponding bills based on logs. When billing disputes occur, the accounts are reconciled to determine the correct number of AI application calls, thereby ensuring accurate billing.

[0011] The AI ​​entities include AI application consumers, AI platform service providers, and AI technology suppliers.

[0012] To optimize the above technical solution, the specific measures also include:

[0013] The specific interaction process between the different AI entities mentioned above is as follows:

[0014] Step 1: AI application consumers purchase AI application services from AI platform service providers. Each time an AI application service is called, the AI ​​application consumer generates a token 1, i.e., Token1, based on asymmetric encryption technology. Then, the AI ​​application service identifier, Token1, and AI application consumer identifier are sent to the AI ​​platform service provider. At the same time, the AI ​​application consumer records a log 1 containing Token1 locally.

[0015] Step 2: After receiving the request from the AI ​​application consumer, the AI ​​platform service provider generates token 2, i.e., Token2, based on asymmetric encryption technology. Then, it sends the AI ​​application service identifier, Token1, and Token2 to the AI ​​technology supplier. At the same time, the AI ​​platform service provider records log 2 containing Token1 and Token2 locally.

[0016] Step 3: After receiving the request from the AI ​​platform service provider, the AI ​​technology provider decrypts Token2 and generates Token3 based on asymmetric encryption technology. Then, Token1, Token2 and Token3 are sent to the AI ​​application service.

[0017] Step 4: The AI ​​application service locally saves Token1, Token2, and Token3. If the processing is normal, it sends a correct response to the AI ​​technology provider; otherwise, it sends an error response.

[0018] Step 5: When the AI ​​technology provider receives a correct response from the AI ​​application service, it returns Token3 to the AI ​​platform service provider and records Log 3 containing Token1, Token2, and Token3 locally.

[0019] Step 6: The AI ​​platform service provider forwards Token1, Token2, and Token3 to the AI ​​application consumer and decrypts Token3. At the same time, it records log 4 containing Token1, Token2, and Token3 locally.

[0020] Step 7: The AI ​​application consumer sends a call request with Token1, Token2 and Token3 to the AI ​​application service. The AI ​​application service matches Token1, Token2 and Token3 in the call request with the three tokens stored locally. If they match completely, the request is processed normally, and a log containing Token1, Token2 and Token3 and the processing result of the request is recorded locally.

[0021] Step 1 above specifically includes:

[0022] AI application consumers purchase AI application services from AI platform service providers;

[0023] Each subsequent use of the AI ​​application service requires the AI ​​application consumer to request a token from the AI ​​platform service provider.

[0024] AI application consumers and AI platform service providers each store the private key and public key of key pair 1 respectively;

[0025] AI application consumers encrypt their own identifier and a random number they generate using a private key to form token 1, or Token1. They then send the AI ​​application service identifier, Token1, and AI application consumer identifier to the AI ​​platform service provider.

[0026] The AI ​​application consumer records Log1 locally. The contents of Log1 include: AI application service identifier, Token1, AI application consumer identifier, and generated random number.

[0027] Step 2 above specifically includes:

[0028] The AI ​​platform service provider and the AI ​​technology supplier each keep the private key and public key of key pair 2 respectively;

[0029] After receiving a request from an AI application consumer, the AI ​​platform service provider encrypts its own identifier and a random number it generates using its private key to form Token2. Then, it sends the AI ​​application service identifier, Token1, and Token2 to the AI ​​technology supplier.

[0030] The AI ​​platform service provider records Log 2 locally. The contents of Log 2 include: AI application service identifier, Token1, AI application consumer identifier and its generated random number, Token2, AI platform service provider identifier and its generated random number.

[0031] Step 3 above specifically includes:

[0032] After receiving the request from the AI ​​platform service provider, the AI ​​technology provider uses the public key in key pair 2 to decrypt Token2 and obtain the AI ​​platform service provider's identifier and its generated random number.

[0033] They then use their private key to encrypt their own identifier and a randomly generated number to form Token3. Subsequently, the AI ​​technology provider sends Token1, Token2, and Token3 to the AI ​​application service.

[0034] Step 5 above specifically includes:

[0035] When the AI ​​technology provider receives a correct response from the AI ​​application service provider, it returns Token3 to the AI ​​platform service provider. At the same time, the AI ​​technology provider records Log3 locally. The contents of Log3 include: AI application service identifier, AI application consumer identifier, Token1, AI platform service provider identifier and its generated random number, Token2, AI platform service provider identifier and its generated random number, Token3, AI technology provider identifier and its generated random number.

[0036] Step 6 above specifically includes:

[0037] The AI ​​platform service provider forwards Token1, Token2, and Token3 to the AI ​​application consumer. At the same time, it uses the public key in Token2 to decrypt Token3 and obtain the AI ​​technology provider's identifier and random number.

[0038] The AI ​​platform service provider records Log 4 locally. The contents of Log 4 include: AI application service identifier, Token1, Token2, Token3, the identifier of the AI ​​technology provider and its generated random number.

[0039] Step 7 above specifically includes:

[0040] AI application consumers initiate calls to AI application services, with Token1, Token2 and Token3 included in the request;

[0041] The AI ​​application service matches the three tokens stored locally. If they match completely, the request is processed normally. After responding normally to the AI ​​application consumer, the three tokens are set to invalid. If the three tokens are received repeatedly in the future, the service will be refused.

[0042] The AI ​​platform service provider logs 5 locally. The contents of log 5 include: AI application service identifier, Token1, Token2, Token3, and the processing result of the request.

[0043] The billing methods for the different AI entities mentioned above are as follows:

[0044] 1) AI Application Consumer: Based on Log 1 and Log 5, a bill record 1 is formed using the AI ​​application service identifier and Token1 as indexes. The content includes: AI application service identifier, Token1, AI application consumer identifier and its generated random number, Token2, Token3, and the processing result of the request.

[0045] 2) AI Platform Service Provider: Based on Log 2 and Log 4, a billing record 2 is formed using the AI ​​application service identifier, Token1, and Token2 as indexes. The content includes: AI application service identifier, Token1, AI application consumer identifier and its generated random number, Token2, AI platform service provider identifier and its generated random number, Token3, and the processing result of the AI ​​application service request.

[0046] 3) AI technology provider: Based on log 3, a bill record 3 is generated, which includes: AI application service identifier, AI application consumer identifier, Token1, AI platform service provider identifier and its generated random number, Token2, AI platform service provider identifier and its generated random number, Token3, AI technology provider identifier and its generated random number.

[0047] The reconciliation method in case of billing disputes is as follows:

[0048] 1) For Token1, AI application consumer identifier and its generated random number in the AI ​​platform service provider's bill, the AI ​​application consumer traverses its own bill. If they exist, it means that the records of both parties are consistent.

[0049] If there is no matching Token1, use the key stored in key pair 1, i.e. the private key in key pair 1, to encrypt the AI ​​application consumer identifier and the random number it generates. If the resulting Token1 does not match the value of Token1 in the AI ​​platform service provider's bill, it means that the Token1 in the AI ​​platform service provider's bill is forged, and the bill is rejected.

[0050] If the obtained Token1 matches the value of Token1 in the AI ​​platform service provider's bill, it means that the AI ​​application consumer billing record is incomplete and the call was missed.

[0051] 2) For Token2, the identifier of the AI ​​platform service provider and its generated random number in the AI ​​technology provider's bill, the AI ​​platform service provider traverses its own bill. If they are found, it means that the records of both parties are consistent.

[0052] If there is no matching Token2, use the key stored in key pair 2 (i.e., the private key in key pair 2) to encrypt the AI ​​platform service provider's identifier and the generated random number. If the resulting Token2 does not match the value of Token2 in the AI ​​technology provider's invoice, it means that the Token2 in the AI ​​technology provider's invoice is forged, and the invoice is rejected.

[0053] If the obtained Token2 matches the value of Token2 in the AI ​​technology provider's bill, it means that the AI ​​platform service provider's billing record is incomplete and that this call was missed.

[0054] 3) AI application consumers, AI platform service providers, and AI technology suppliers conduct regular reconciliation operations. After each reconciliation, key pair 1 and key pair 2 are regenerated and stored by each of the three parties.

[0055] The present invention has the following beneficial effects:

[0056] This invention designs a process for invoking AI applications, recording logs for interactions between different entities. These logs include token information using asymmetric encryption technology. Even if the plaintext is known, the key remains unknown, making the token information impossible to forge. Therefore, in case of billing disputes, log entries can be compared one by one to accurately count the correct number of AI application calls. The call counts calculated by all three parties are neither missed nor over-counted, achieving token-based reconciliation to prevent any party from forging or intentionally omitting AI call counts. Attached Figure Description

[0057] Figure 1 The call process for AI application services. Detailed Implementation

[0058] The embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

[0059] This invention provides a statistical method for counting AI application calls. During the AI ​​application call process, logs are recorded separately when different AI entities interact, including token information based on asymmetric encryption technology.

[0060] In the event of a billing dispute, log entries are compared one by one to determine the correct number of AI application calls, thus ensuring accurate billing.

[0061] The AI ​​entities include AI application consumers, AI platform service providers, AI technology suppliers, and AI application services.

[0062] In practice, the public key and private key mentioned in this invention refer only to the two keys in the key pair, which are kept by the two communicating parties respectively and are not disclosed.

[0063] AI application services are services developed, deployed, and maintained by AI technology providers and belong to the same party.

[0064] like Figure 1 As shown, the specific interaction details are as follows:

[0065] Step 1: The AI ​​application consumer purchases AI application services from the AI ​​platform service provider. Each time the AI ​​application service is invoked, the AI ​​application consumer generates a token 1 (Token1) based on asymmetric encryption technology. Then, it sends the AI ​​application service identifier, Token1, and the AI ​​application consumer identifier to the AI ​​platform service provider. Simultaneously, the AI ​​application consumer records log 1 containing Token1 locally. Specifically:

[0066] AI application consumers purchase AI application services from AI platform service providers;

[0067] Each time the AI ​​application service is called subsequently, the AI ​​application consumer will apply for a token from the AI ​​platform service provider;

[0068] AI application consumers and AI platform service providers each store the private key and public key of key pair 1 respectively;

[0069] AI application consumers encrypt their own identifier and a random number they generate using a private key to form token 1, or Token1. They then send the AI ​​application service identifier, Token1, and AI application consumer identifier to the AI ​​platform service provider.

[0070] The AI ​​application consumer records Log1 locally. The contents of Log1 include: AI application service identifier, Token1, AI application consumer identifier, and generated random number.

[0071] Step 2: After receiving the request from the AI ​​application consumer, the AI ​​platform service provider generates token 2 (Token2) based on asymmetric encryption technology. It then sends the AI ​​application service identifier, token 1, and token 2 to the AI ​​technology supplier. Simultaneously, the AI ​​platform service provider locally records log 2 containing token 1 and token 2. Specifically:

[0072] The AI ​​platform service provider and the AI ​​technology supplier each keep the private key and public key of key pair 2 respectively;

[0073] After receiving a request from an AI application consumer, the AI ​​platform service provider encrypts its own identifier and a random number it generates using its private key to form Token2. Then, it sends the AI ​​application service identifier, Token1, and Token2 to the AI ​​technology supplier.

[0074] The AI ​​platform service provider records Log 2 locally. The contents of Log 2 include: AI application service identifier, Token1, AI application consumer identifier and its generated random number, Token2, AI platform service provider identifier and its generated random number.

[0075] Step 3: After receiving the request from the AI ​​platform service provider, the AI ​​technology provider decrypts Token2 and generates Token3 using asymmetric encryption technology. Then, Token1, Token2, and Token3 are sent to the AI ​​application service. Specifically:

[0076] After receiving a request from the AI ​​platform service provider, the AI ​​technology provider uses the public key in key pair 2 to decrypt Token2 and obtain the AI ​​platform service provider's identifier and its generated random number.

[0077] They then use their private key to encrypt their own identifier and a self-generated random number to form Token3. Subsequently, the AI ​​technology provider sends Token1, Token2, and Token3 to the AI ​​application service.

[0078] Step 4: The AI ​​application service locally saves Token1, Token2, and Token3. If the processing is normal, it sends a correct response to the AI ​​technology provider; otherwise, it sends an error response.

[0079] Step 5: When the AI ​​technology provider receives a correct response from the AI ​​application service provider, it returns Token3 to the AI ​​platform service provider and simultaneously records Log 3 containing Token1, Token2, and Token3 locally; specifically:

[0080] When the AI ​​technology provider receives a correct response from the AI ​​application service provider, it returns Token3 to the AI ​​platform service provider. At the same time, the AI ​​technology provider records Log3 locally. The contents of Log3 include: AI application service identifier, AI application consumer identifier, Token1, AI platform service provider identifier and its generated random number, Token2, AI platform service provider identifier and its generated random number, Token3, AI technology provider identifier and its generated random number.

[0081] Step 6: The AI ​​platform service provider forwards Token1, Token2, and Token3 to the AI ​​application consumer and decrypts Token3. Simultaneously, it records log 4 containing Token1, Token2, and Token3 locally. Specifically:

[0082] The AI ​​platform service provider forwards Token1, Token2, and Token3 to AI application consumers. Simultaneously, using the public key from Token2, they decrypt Token3 to obtain the AI ​​technology provider's identifier and a random number.

[0083] The AI ​​platform service provider records Log 4 locally. The contents of Log 4 include: AI application service identifier, Token1, Token2, Token3, the identifier of the AI ​​technology provider and its generated random number.

[0084] Step 7: The AI ​​application consumer sends a call request with Token1, Token2, and Token3 to the AI ​​application service. The AI ​​application service matches Token1, Token2, and Token3 in the call request with three locally stored tokens. If there is a complete match, the request is processed normally, and a log containing Token1, Token2, Token3, and the processing result of the request is recorded locally. Specifically:

[0085] AI application consumers initiate calls to AI application services, with Token1, Token2, and Token3 included in the request.

[0086] The AI ​​application service matches the three tokens stored locally. If they match completely, the request is processed normally. After responding normally to the AI ​​application consumer, the three tokens are set to invalid. If the three tokens are received repeatedly in the future, the service will be refused.

[0087] The AI ​​platform service provider logs 5 locally. The contents of log 5 include: AI application service identifier, Token1, Token2, Token3, and the processing result of the request (success or failure).

[0088] The formation of the bill:

[0089] 1. Based on logs 1 and 5, the AI ​​application consumer creates a bill record 1 indexed by "AI application service identifier" and "Token1". The record includes: AI application service identifier, Token1, AI application consumer identifier and its generated random number, Token2, Token3, and the processing result of the request (success or failure).

[0090] 2. Based on logs 2 and 4, the AI ​​platform service provider creates a billing record 2 indexed by "AI application service identifier", "Token1" and "Token2". The record includes: AI application service identifier, Token1, AI application consumer identifier and its generated random number, Token2, AI platform service provider identifier and its generated random number, Token3, and the processing result of the AI ​​application service request.

[0091] 3. Based on log 3, the AI ​​technology provider generates a billing record 3, which includes: AI application service identifier, AI application consumer identifier, Token1, AI platform service provider identifier and its generated random number, Token2, AI platform service provider identifier and its generated random number, Token3, AI technology provider identifier and its generated random number.

[0092] Reconciliation methods when billing issues occur:

[0093] 1. Regarding "Token1, AI application consumer identifier and its generated random number" in the AI ​​platform service provider's bill, the AI ​​application consumer can iterate through their own bill. If they exist, it means that the records of both parties are consistent.

[0094] If a matching Token1 is not available, the private key stored in key pair 1 can be used to encrypt the "AI application consumer identifier and its generated random number." If the resulting Token1 does not match the value of Token1 in the AI ​​platform service provider's bill, it indicates that the Token1 in the AI ​​platform service provider's bill is forged, and the bill can be rejected. This reconciliation method can prevent AI platform service providers from artificially inflating the number of times AI applications are called.

[0095] If the obtained Token1 matches the value of Token1 in the AI ​​platform service provider's bill, it indicates that the AI ​​application consumer billing record is incomplete and that the call was missed. This reconciliation method can prevent the AI ​​application from missing calls in the AI ​​application consumer billing.

[0096] 2. Regarding "Token2, AI platform service provider's identifier and its generated random number" in the AI ​​technology provider's bill, the AI ​​platform service provider can iterate through its own bills. If they exist, it means that the records of both parties are consistent.

[0097] If a matching Token2 is not available, the private key stored in key pair 2 can be used to encrypt the "AI platform service provider's identifier and its generated random number." If the resulting Token2 does not match the value of Token2 in the AI ​​technology provider's invoice, it indicates that the Token2 in the AI ​​technology provider's invoice is forged, and the invoice can be rejected. This reconciliation method can prevent AI technology providers from artificially inflating the number of times AI applications are called.

[0098] If the resulting Token2 matches the value of Token2 in the AI ​​technology provider's invoice, it indicates that the AI ​​platform service provider's invoice record is incomplete and that the call was missed. This reconciliation method can prevent the AI ​​application calls from being omitted from the AI ​​platform service provider's invoice.

[0099] 3. The three parties shall conduct reconciliation operations regularly. After each reconciliation, key pair 1 and key pair 2 shall be regenerated and each party shall keep the key.

[0100] Specifically, this invention can use asymmetric encryption algorithms such as AES.

[0101] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principle of the present invention should be considered within the scope of protection of the present invention.

Claims

1. A statistical method for counting AI application calls, characterized in that, include: During the AI ​​application invocation process, logs are recorded separately when different AI entities interact, including token information based on asymmetric encryption technology; Different AI entities generate corresponding bills based on logs. When billing disputes occur, the accounts are reconciled to determine the correct number of AI application calls, thereby ensuring accurate billing. The AI ​​entities include AI application consumers, AI platform service providers, and AI technology suppliers; The specific interaction process between different AI entities is as follows: Step 1: AI application consumers purchase AI application services from AI platform service providers. Each time an AI application service is called, the AI ​​application consumer generates a token 1, i.e., Token1, based on asymmetric encryption technology. Then, the AI ​​application service identifier, Token1, and AI application consumer identifier are sent to the AI ​​platform service provider. At the same time, the AI ​​application consumer records a log 1 containing Token1 locally. Step 2: After receiving the request from the AI ​​application consumer, the AI ​​platform service provider generates token 2, i.e., Token2, based on asymmetric encryption technology. Then, it sends the AI ​​application service identifier, Token1, and Token2 to the AI ​​technology supplier. At the same time, the AI ​​platform service provider records log 2 containing Token1 and Token2 locally. Step 3: After receiving the request from the AI ​​platform service provider, the AI ​​technology provider decrypts Token2 and generates Token3 based on asymmetric encryption technology. Then, Token1, Token2 and Token3 are sent to the AI ​​application service. Step 4: The AI ​​application service locally saves Token1, Token2, and Token3. If the processing is normal, it sends a correct response to the AI ​​technology provider; otherwise, it sends an error response. Step 5: When the AI ​​technology provider receives a correct response from the AI ​​application service, it returns Token3 to the AI ​​platform service provider and records Log 3 containing Token1, Token2, and Token3 locally. Step 6: The AI ​​platform service provider forwards Token1, Token2, and Token3 to the AI ​​application consumer and decrypts Token3. At the same time, it records log 4 containing Token1, Token2, and Token3 locally. Step 7: The AI ​​application consumer sends a call request with Token1, Token2 and Token3 to the AI ​​application service. The AI ​​application service matches Token1, Token2 and Token3 in the call request with the three tokens stored locally. If they match completely, the request is processed normally, and a log 5 containing Token1, Token2 and Token3 and the processing result of the request is recorded locally.

2. The statistical method for counting AI application calls according to claim 1, characterized in that, Step 1 specifically includes: AI application consumers purchase AI application services from AI platform service providers; Each subsequent use of the AI ​​application service requires the AI ​​application consumer to request a token from the AI ​​platform service provider. AI application consumers and AI platform service providers each store the private key and public key of key pair 1 respectively; AI application consumers encrypt their own identifier and a random number they generate using a private key to form token 1, or Token1. They then send the AI ​​application service identifier, Token1, and AI application consumer identifier to the AI ​​platform service provider. The AI ​​application consumer records Log1 locally. The contents of Log1 include: AI application service identifier, Token1, AI application consumer identifier, and generated random number.

3. The statistical method for counting AI application calls according to claim 1, characterized in that, Step 2 specifically includes: The AI ​​platform service provider and the AI ​​technology supplier each keep the private key and public key of key pair 2 respectively; After receiving a request from an AI application consumer, the AI ​​platform service provider encrypts its own identifier and a random number it generates using its private key to form Token2. Then, it sends the AI ​​application service identifier, Token1, and Token2 to the AI ​​technology supplier. The AI ​​platform service provider records Log 2 locally. The contents of Log 2 include: AI application service identifier, Token1, AI application consumer identifier and its generated random number, Token2, AI platform service provider identifier and its generated random number.

4. The statistical method for counting AI application calls according to claim 1, characterized in that, Step 3 specifically includes: After receiving the request from the AI ​​platform service provider, the AI ​​technology provider uses the public key in key pair 2 to decrypt Token2 and obtain the AI ​​platform service provider's identifier and its generated random number. They then use their private key to encrypt their own identifier and a random number they generate to form Token3. Subsequently, the AI ​​technology provider sends Token1, Token2, and Token3 to the AI ​​application service.

5. The statistical method for counting AI application calls according to claim 1, characterized in that, Step 5 specifically includes: When the AI ​​technology provider receives a correct response from the AI ​​application service provider, it returns Token3 to the AI ​​platform service provider. At the same time, the AI ​​technology provider records Log3 locally. The contents of Log3 include: AI application service identifier, AI application consumer identifier, Token1, AI platform service provider identifier and its generated random number, Token2, AI platform service provider identifier and its generated random number, Token3, AI technology provider identifier and its generated random number.

6. The statistical method for counting AI application calls according to claim 1, characterized in that, Step 6 specifically includes: The AI ​​platform service provider forwards Token1, Token2, and Token3 to the AI ​​application consumer. At the same time, it uses the public key in Token2 to decrypt Token3 and obtain the AI ​​technology provider's identifier and random number. The AI ​​platform service provider records Log 4 locally. The contents of Log 4 include: AI application service identifier, Token1, Token2, Token3, the identifier of the AI ​​technology provider and its generated random number.

7. The statistical method for counting AI application calls according to claim 1, characterized in that, Step 7 specifically includes: AI application consumers initiate calls to AI application services, with Token1, Token2 and Token3 included in the request; The AI ​​application service matches the three tokens stored locally. If they match completely, the request is processed normally. After responding normally to the AI ​​application consumer, the three tokens are set to invalid. If the three tokens are received repeatedly in the future, the service will be refused. The AI ​​platform service provider logs 5 locally. The contents of log 5 include: AI application service identifier, Token1, Token2, Token3, and the processing result of the request.

8. The statistical method for counting AI application calls according to claim 1, characterized in that, The billing methods for different AI entities are as follows: 1) AI Application Consumer: Based on Log 1 and Log 5, a bill record 1 is formed using the AI ​​application service identifier and Token1 as indexes. The content includes: AI application service identifier, Token1, AI application consumer identifier and its generated random number, Token2, Token3, and the processing result of the request. 2) AI Platform Service Provider: Based on Log 2 and Log 4, a billing record 2 is formed using the AI ​​application service identifier, Token1, and Token2 as indexes. The content includes: AI application service identifier, Token1, AI application consumer identifier and its generated random number, Token2, AI platform service provider identifier and its generated random number, Token3, and the processing result of the AI ​​application service request. 3) AI technology provider: Based on log 3, a bill record 3 is generated, which includes: AI application service identifier, AI application consumer identifier, Token1, AI platform service provider identifier and its generated random number, Token2, AI platform service provider identifier and its generated random number, Token3, AI technology provider identifier and its generated random number.

9. The statistical method for counting AI application calls according to claim 8, characterized in that, The reconciliation method in case of billing disputes is as follows: 1) For Token1, AI application consumer identifier and its generated random number in the AI ​​platform service provider's bill, the AI ​​application consumer traverses its own bill. If they exist, it means that the records of both parties are consistent. If there is no matching Token1, use the key stored in key pair 1, i.e. the private key in key pair 1, to encrypt the AI ​​application consumer identifier and the random number it generates. If the resulting Token1 does not match the value of Token1 in the AI ​​platform service provider's bill, it means that the Token1 in the AI ​​platform service provider's bill is forged, and the bill is rejected. If the obtained Token1 matches the value of Token1 in the AI ​​platform service provider's bill, it means that the AI ​​application consumer billing record is incomplete and the call was missed. 2) For Token2, the identifier of the AI ​​platform service provider and its generated random number in the AI ​​technology provider's bill, the AI ​​platform service provider traverses its own bill. If they are found, it means that the records of both parties are consistent. If there is no matching Token2, use the key stored in key pair 2 (i.e., the private key in key pair 2) to encrypt the AI ​​platform service provider's identifier and the generated random number. If the resulting Token2 does not match the value of Token2 in the AI ​​technology provider's invoice, it means that the Token2 in the AI ​​technology provider's invoice is forged, and the invoice is rejected. If the obtained Token2 matches the value of Token2 in the AI ​​technology provider's bill, it means that the AI ​​platform service provider's billing record is incomplete and that this call was missed. 3) AI application consumers, AI platform service providers, and AI technology suppliers conduct regular reconciliation operations. After each reconciliation, key pair 1 and key pair 2 are regenerated and stored by each of the three parties.

Images