[KongchangAI]
Model

SimRec

A reinforcement-learning-based training framework for recommendation agents. It treats the complete interaction policy of a tool-integrated recommendation agent, including search, follow-up questioning, and re-recommendation, as the optimization target, employs a task-anchored user simulator and task-aligned rewards, and uses GRPO to optimize the entire trajectory.

Timeline (last 90 days)

Nov 13

SimRec is a reinforcement learning framework whose goal is to train the complete interaction policy of a tool-integrated recommendation agent, rather than separately tuning the search, ranking, and follow-up questioning modules.

Unverified50%
Nov 13

SimRec introduces a task-anchored user simulator that does not rely on user profiles but instead conditions its feedback on the current task requirements.

Unverified50%
Nov 13

SimRec's task-aligned reward uses GRPO to optimize the entire trajectory, with rewards including the final hit result, natural retrieval quality, and effective tool use.

Unverified50%
Nov 13

SimRec employs a training-only goal injection mechanism to alleviate the sparse reward problem.

Unverified50%
Nov 13

Goal injection works by providing the agent or simulator with additional partial information about the target item during training (such as category and attribute keywords), artificially increasing the reachability of the task, and it is used only during training and completely removed at inference.

Unverified50%
Nov 13

On the ESCI dataset, SimRec raises a Qwen3-4B agent's NDCG@100 from 32.23% to 35.42%.

Unverified50%
Nov 13

The trained 4B model surpasses much larger training-free agents in NDCG and also outperforms task-aligned baseline methods.

Unverified50%
Nov 13

The trained agent is noticeably more frequent in three behaviors: verifying candidate items, leveraging user preferences, and recovering from rejected recommendations to offer a new plan.

Unverified50%
Nov 13

The SimRec code is available at https://github.com/ZJU-REAL/SimRec.

Unverified50%
Nov 13

The gap between the simulator and real users, as well as the feedback quality in self-play mode, remain aspects of this type of method that require continuous validation in deployment.

Unverified50%

All Facts (10)

Unverified

The gap between the simulator and real users, as well as the feedback quality in self-play mode, remain aspects of this type of method that require continuous validation in deployment.

50%
Unverified

SimRec introduces a task-anchored user simulator that does not rely on user profiles but instead conditions its feedback on the current task requirements.

50%
Unverified

SimRec's task-aligned reward uses GRPO to optimize the entire trajectory, with rewards including the final hit result, natural retrieval quality, and effective tool use.

50%
Unverified

SimRec employs a training-only goal injection mechanism to alleviate the sparse reward problem.

50%
Unverified

Goal injection works by providing the agent or simulator with additional partial information about the target item during training (such as category and attribute keywords), artificially increasing the reachability of the task, and it is used only during training and completely removed at inference.

50%
Unverified

On the ESCI dataset, SimRec raises a Qwen3-4B agent's NDCG@100 from 32.23% to 35.42%.

50%
Unverified

The trained 4B model surpasses much larger training-free agents in NDCG and also outperforms task-aligned baseline methods.

50%
Unverified

The trained agent is noticeably more frequent in three behaviors: verifying candidate items, leveraging user preferences, and recovering from rejected recommendations to offer a new plan.

50%
Unverified

SimRec is a reinforcement learning framework whose goal is to train the complete interaction policy of a tool-integrated recommendation agent, rather than separately tuning the search, ranking, and follow-up questioning modules.

50%
Unverified

The SimRec code is available at https://github.com/ZJU-REAL/SimRec.

50%

Source Articles