🤖 AI Summary
This work addresses the limitations of existing generative recommendation methods based on semantic IDs, which fail to effectively exploit their hierarchical structure, intra-layer relationships, and item neighborhoods to construct an explicit reasoning space, and exhibit insufficient coupling between natural language reasoning and the discrete semantic ID space. To bridge this gap, we propose CogRec, a novel framework that, for the first time, embeds structural cognitive reasoning into the semantic ID generation space. CogRec constructs structured reasoning paths through vertical hierarchy, intra-layer semantic graphs, and item neighborhoods, and introduces a Semantic ID (SID) Routing mechanism supporting match, lateral jump, and exploration operations. Multi-stage supervised training under trie constraints aligns language generation with the SID routing branch. Experiments on three sequential recommendation benchmarks show that SID Routing significantly outperforms direct generation; structured reasoning achieves optimal performance when prefix matching is weak but learnable transitions exist, whereas overly long or poorly supported paths introduce additional decoding overhead and errors.
📝 Abstract
Semantic-ID-based generative recommendation represents each item as a hierarchical discrete token sequence and reformulates next-item prediction as constrained sequence generation. Existing methods, however, mainly use Semantic IDs as target sequences to be memorized, leaving the hierarchy, intra-layer relations, and item neighborhoods underused as an explicit reasoning space. Explicit reasoning-enhanced generative methods often produce a natural-language rationale before the item identifier, but this rationale is only weakly coupled with the discrete SID space in which the final prediction is made. We propose CogRec, a structure-cognitive fast-and-slow reasoning framework that grounds intermediate reasoning in the same SID topology used for target generation. CogRec augments the vertical SID hierarchy with intra-layer semantic graphs and item-level neighborhoods, and introduces SID Routing to represent recommendation reasoning through layer-wise Match, LateralJump, and Explore operations. Exact matching implements fast semantic localization, whereas lateral and exploratory operations instantiate slower structural navigation. A supervised multi-stage pipeline aligns the newly introduced SID tokens, establishes direct SID generation, and trains natural-language and SID-routing reasoning branches from a shared checkpoint under the same trie-constrained output space. Experiments on three public sequential-recommendation benchmarks show that SID Routing improves its corresponding direct-generation, indicate that structure-grounded reasoning is most useful when prefix matching is insufficient but learnable SID-space transitions remain available, whereas long or weakly supported routes introduce additional decoding cost and accumulated errors. Code is available at https://github.com/caskcsg/CogRec