Method, apparatus for managing recommendation policy
A method, apparatus, electronic device, storage medium and program product of code management are provided. In response to a request for building an executable file, corresponding developed code is obtained from a code library. The developed code is compiled into intermediate code to determine security of the intermediate code. In response to determining that the intermediate code is secure, an executable file is generated based on the intermediate code.
1. A method for managing a recommendation policy in a target application, comprising:
obtaining, by a recommendation management sub-system, a group of object features associated with a group of objects in the target application, the group of object features being converted from attributes of the group of objects, wherein the group of objects do not directly characterize the attributes of the group of objects;
determining, by the recommendation management sub-system, a first object feature and a second object feature from the group of object features, a first difference between the first object feature and the second object feature being less than a first threshold;
determining, by the recommendation management sub-system, a first recommendation result corresponding to the first object feature and a second recommendation result corresponding to the second object feature based on the recommendation policy in the target application; and
evaluating, by the recommendation management sub-system, the recommendation policy based on the first recommendation result and the second recommendation result, comprising:
determining, by the recommendation management sub-system, a second difference between the first recommendation result and the second recommendation result; and
determining, by the recommendation management sub-system, fairness of the recommendation policy based on a comparison between the second difference and a second threshold,
wherein determining the first recommendation result corresponding to the first object feature and the second recommendation result corresponding to the second object feature comprises:
transmitting the first object feature to a recommendation model associated with the recommendation policy to determine the first recommendation result, the recommendation model operating remotely via an API provided by the target application; and
providing the second object feature to the recommendation model to determine the second recommendation result.
2. The method according to claim 1 , wherein obtaining the group of object features associated with the group of objects in the target application comprises:
obtaining the group of object features via an application program interface API provided by the target application.
3. The method according to claim 1 , wherein the first recommendation result and the second recommendation result are vector representations output by the recommendation model.
4. The method according to claim 1 , further comprising:
obtaining a third object feature and a historical recommendation result for the third object feature from the target application; and
determining fairness of the recommendation policy based on correlation between the third object feature and the historical recommendation result.
5. The method according to claim 1 , wherein the recommendation policy is used to recommend at least one multimedia content to a user in the target application.
6. The method according to claim 1 , further comprising:
obtaining source code corresponding to the recommendation policy; and
evaluating the recommendation policy based on the source code or intermediate code corresponding to the source code.
7. An electronic device, comprising:
a memory and a processor;
wherein the memory is used to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor to implement a method for managing a recommendation policy in a target application, comprising:
obtaining, by a recommendation management sub-system, a group of object features associated with a group of objects in the target application, the group of object features being converted from attributes of the group of objects, wherein the group of objects do not directly characterize the attributes of the group of objects;
determining, by the recommendation management sub-system, a first object feature and a second object feature from the group of object features, a first difference between the first object feature and the second object feature being less than a first threshold;
determining, by the recommendation management sub-system, a first recommendation result corresponding to the first object feature and a second recommendation result corresponding to the second object feature based on the recommendation policy in the target application; and
evaluating, by the recommendation management sub-system, the recommendation policy based on the first recommendation result and the second recommendation result, comprising:
determining, by the recommendation management sub-system, a second difference between the first recommendation result and the second recommendation result; and
determining, by the recommendation management sub-system, fairness of the recommendation policy based on a comparison between the second difference and a second threshold,
wherein determining the first recommendation result corresponding to the first object feature and the second recommendation result corresponding to the second object feature comprises:
transmitting the first object feature to a recommendation model associated with the recommendation policy to determine the first recommendation result, the recommendation model operating remotely via an API provided by the target application; and
providing the second object feature to the recommendation model to determine the second recommendation result.
8. A non-transitory computer-readable storage medium, with one or more computer instructions stored thereon, wherein the one or more computer instructions are executed by a processor to implement a method for managing a recommendation policy in a target application, comprising:
obtaining, by a recommendation management sub-system, a group of object features associated with a group of objects in the target application, the group of object features being converted from attributes of the group of objects, wherein the group of objects do not directly characterize the attributes of the group of objects;
determining, by the recommendation management sub-system, a first object feature and a second object feature from the group of object features, a first difference between the first object feature and the second object feature being less than a first threshold;
determining, by the recommendation management sub-system, a first recommendation result corresponding to the first object feature and a second recommendation result corresponding to the second object feature based on the recommendation policy in the target application; and
evaluating, by the recommendation management sub-system, the recommendation policy based on the first recommendation result and the second recommendation result, comprising:
determining, by the recommendation management sub-system, a second difference between the first recommendation result and the second recommendation result; and
determining, by the recommendation management sub-system, fairness of the recommendation policy based on a comparison between the second difference and a second threshold,
wherein determining the first recommendation result corresponding to the first object feature and the second recommendation result corresponding to the second object feature comprises:
transmitting the first object feature to a recommendation model associated with the recommendation policy to determine the first recommendation result, the recommendation model operating remotely via an API provided by the target application; and
providing the second object feature to the recommendation model to determine the second recommendation result.
9. A computer program product, embodied on a non-transitory computer readable medium and comprising one or more computer instructions, wherein the one or more computer instructions are executed by a processor to perform actions of:
obtaining, by a recommendation management sub-system, a group of object features associated with a group of objects in a target application, the group of object features being converted from attributes of the group of objects, wherein the group of objects do not directly characterize the attributes of the group of objects;
determining, by the recommendation management sub-system, a first object feature and a second object feature from the group of object features, a first difference between the first object feature and the second object feature being less than a first threshold;
determining, by the recommendation management sub-system, a first recommendation result corresponding to the first object feature and a second recommendation result corresponding to the second object feature based on the recommendation policy in the target application; and
evaluating, by the recommendation management sub-system, the recommendation policy based on the first recommendation result and the second recommendation result, comprising:
determining, by the recommendation management sub-system, a second difference between the first recommendation result and the second recommendation result; and
determining, by the recommendation management sub-system, fairness of the recommendation policy based on a comparison between the second difference and a second threshold,
wherein determining the first recommendation result corresponding to the first object feature and the second recommendation result corresponding to the second object feature comprises:
transmitting the first object feature to a recommendation model associated with the recommendation policy to determine the first recommendation result, the recommendation model operating remotely via an API provided by the target application; and
providing the second object feature to the recommendation model to determine the second recommendation result.