Service execution system and related product

1 . A system for service execution comprising:

a service execution apparatus configured to:

collect a first quantity of service requests from a first received service request to obtain a first sample set;

send the first sample set;

receive a first model or a first policy; and

process a second received service request based on one of the first model or the first policy; and

a decision apparatus configured to:

receive the first sample set from the service execution apparatus;

evaluate the first model and the first policy based on the first sample set to obtain a first prediction accuracy of the first model and a second prediction accuracy of the first policy;

send the first model to the service execution apparatus when the first prediction accuracy is greater than the second prediction accuracy; and

send the first policy to the service execution apparatus when the second prediction accuracy is greater than the first prediction accuracy.

2 . The system according to claim 1 , wherein the service execution apparatus is further configured to:

receive a historical service request;

collect a second quantity of service requests from the first received service request to obtain a second sample set, wherein the second sample set comprises a training sample set and a test sample set; and

send the second sample set and the historical service request to the decision apparatus.

3 . The system according to claim 2 , wherein the decision apparatus is further configured to:

obtain a plurality of policies; and

evaluate the plurality of policies based on the second sample set to select the first policy, wherein the first policy has a highest prediction accuracy in the plurality of policies.

4 . The system according to claim 2 , wherein the first sample set and the second sample set are similar sample sets, or wherein the first sample set comprises the second sample set.

5 . A method for decision making the method comprising:

receiving, from a service execution apparatus, a first sample set, wherein the first sample set comprises a first quantity of service requests from a first received service request;

evaluating a first model and a first policy based on the first sample set to obtain a first prediction accuracy of the first model and a second prediction accuracy of the first policy;

sending the first model to the service execution apparatus when the first prediction accuracy is greater than the second prediction accuracy; and

sending the first policy to the service execution apparatus when the second prediction accuracy is greater than the first prediction accuracy.

6 . The method according to claim 5 , further comprising:

receiving a second sample set, wherein the second sample set comprises a second quantity of service requests from the first received service request, and wherein the second sample set comprises a training sample set and a test sample set;

obtaining a plurality of models, wherein the plurality of models comprise a second model based on the training sample set and a third model based on a historical service request; and

evaluating the plurality of models based on the test sample set, to select the first model, wherein the first model has a highest prediction accuracy in the plurality of models.

7 . The method according to claim 6 , further comprising:

obtaining a plurality of policies; and

evaluating the plurality of policies based on the second sample set to select the first policy, wherein the first policy is a policy having a best evaluation result in the plurality of policies.

8 . The method according to claim 6 , wherein the first sample set and the second sample set are similar sample sets, or the first sample set comprises the second sample set.

9 . A computing device comprising:

at least one memory configured to store instructions; and

at least one processor coupled to the at least one memory and configured to execute the instructions to cause the computing device to:

obtain, from a service execution apparatus, a first sample set, wherein the first sample set comprises a first quantity of service requests from a first received service request;

evaluate a first model and a first policy based on the first sample set to obtain a first prediction accuracy of the first model and a second prediction accuracy of the first policy;

send the first model to the service execution apparatus when the first prediction accuracy is greater than the second prediction accuracy; and

send the first policy to the service execution apparatus when the second prediction accuracy is greater than the first prediction accuracy.

10 . The system according to claim 2 , wherein the service execution apparatus is further configured to collect the second quantity of service requests before the decision apparatus evaluates the first model and the first policy.

11 . The system according to claim 2 , wherein the decision apparatus is further configured to:

obtain a plurality of models including a second model based on the training sample set and a third model based on the historical service request; and

evaluate the plurality of models based on the test sample set to select the first model from the plurality of models, wherein the first model has a highest prediction accuracy in the plurality of models.

12 . The system according to claim 11 , wherein the decision apparatus is further configured to:

obtain the second model by training an initial model based on the training sample set; and

obtain the third model by training the initial model based on the historical service request.

13 . The system according to claim 11 , wherein the service execution apparatus is further configured to receive the historical service request before collecting the second quantity of service requests from the first received service request.

14 . The system according to claim 3 , wherein the decision apparatus is further configured to obtain the plurality of policies before evaluating the first model and the first policy.

15 . The method according to claim 6 , wherein the first sample set and the second sample set are similar sample sets, or wherein the first sample set comprises the second sample set.

16 . The method according to claim 6 , wherein the second sample set is received before evaluating the first model and the first policy.

17 . The method according to claim 6 , further comprising:

obtaining the second model by training an initial model based on the training sample set; and

obtaining the third model by training the initial model based on the historical service request.

18 . The method according to claim 6 , further comprising receiving, from the service execution apparatus, the historical service request before collecting the second quantity of service requests.

19 . The method according to claim 7 , wherein the plurality of policies is obtained before evaluating the first model and the first policy.

20 . The method according to claim 5 , wherein the first policy is a least recently used (LRU) policy in a cache, and wherein the LRU policy indicates that a least recently accessed object in a cache is evicted to store a new object when storage space of the cache is insufficient.