The exploration of bundle recommendation has garnered significant attention for its potential to enhance user experience and augment business sales. Previous research in this domain has primarily focused on modeling user-item and user-bundle interactions, utilizing multi-view collaboration to bolster the accuracy of bundle recommendations. Nevertheless, existing methodologies exhibit limitations, notably in the inadequate modeling of multi-view information and the absence of multi-grained details. Consequently, addressing the intricate correlation among users, items, and bundles necessitates a sophisticated approach capable of capturing both global and local nuances. We present a novel framework named Large Language Model-based Multi-View and Multi-Grained Learning for Bundle Recommendation (LLMCBR). We introduce an LLM-based semantic refinement module to summarize and encode bundle-level knowledge. To bridge the gap between semantic representation and collaborative signals, we design an adaptation strategy. Furthermore, LLMCBR leverages multi-view and multi-granular modeling to unify collaborative signals. Specifically, LLMCBR integrates item preferences within both bundle-view and item-view, thereby augmenting the comprehensiveness of multi-view data. Following this integration, each view undergoes stratification into multiple granularities to facilitate the acquisition of multi-grained details. We introduce a multiple contrastive instance mechanism to regulate the influence of different granularities and views. This mechanism empowers the model to comprehend complex consumer behaviors across various dimensions. LLMCBR is extensively evaluated over three real-world datasets, and the experimental results demonstrate its superiority.

In real-world cooking scenarios, users often need to create personalized recipes based on limited ingredients, dietary goals, and various restrictions, such as the availability of equipment, flavor preferences, and health conditions. Existing recipe generation methods lack the flexibility and adaptability to meet these individual needs. We present LetMeCook, an end-to-end interactive system for personalized recipe generation that leverages multimodal perception, hybrid retrieval, and content generation. Given a photo of the user's refrigerator and a dietary profile, LetMeCook detects available ingredients, retrieves relevant recipe candidates, and generatively refines them based on user requirements, such as ingredient substitution and flavor adjustment. The system provides both textual and visual previews of the adapted recipes, offering a highly interactive and user-centric experience.

Top-k sparsification has recently been widely used to reduce the communication volume in distributed deep learning. However, due to the Sparse Gradient Accumulation (SGA) dilemma, the performance of top-k sparsification still has limitations. Recently, a few methods have been put forward to handle the SGA dilemma. Regrettably, even the state-of-the-art method suffers from several drawbacks, e.g., it relies on an inefficient communication algorithm and requires extra transmission steps. Motivated by the limitations of existing methods, we propose a novel efficient sparse communication framework, called SparDL. Specifically, SparDL uses the Spar-Reduce-Scatter algorithm, which is based on an efficient Reduce-Scatter model, to handle the SGA dilemma without additional communication operations. Besides, to further reduce the latency cost and improve the efficiency of SparDL, we propose the Spar-All-Gather algorithm. Moreover, we propose the global residual collection algorithm to ensure fast convergence of model training. Finally, extensive experiments are conducted to validate the superiority of SparDL.

Learning item representation is crucial for a myriad of on-line e-commerce applications. The nucleus of retail item representation learning is how to properly fuse the semantics within a single item, and the interactions across different items generated by user behaviors (e.g., co-click or co-view). Product semantics depict the intrinsic characteristics of the item, while the interactions describe the relationships between items from the perspective of human perception. Existing approaches either solely rely on a single type of information or loosely couple them together, leading to hindered representations. In this work, we propose a novel model named TESPA to reinforce semantic modeling and interaction modeling mutually. Specifically, collaborative filtering signals in the interaction graph are encoded into the language models through fine-grained topological pre-training, and the interaction graph is further enriched based on semantic similarities. After that, a novel multi-channel co-training paradigm is proposed to deeply fuse the semantics and interactions under a unified framework. In a nutshell, TESPA is capable of enjoying the merits of both sides to facilitate item representation learning. Experimental results of on-line and off-line evaluations demonstrate the superiority of our proposal.

Materialized model query aims to find the most appropriate materialized model as the initial model for model reuse. It is the precondition of model reuse, and has recently attracted much attention. Nonetheless, the existing methods suffer from the need to provide source data, limited range of applications, and inefficiency since they do not construct a suitable metric to measure the target-related knowledge of materialized models. To address this, we present${\sf MMQ}$, a source-data free, general, efficient, and effective materialized model query framework. It uses a Gaussian mixture-based metric called separation degree to rank materialized models. For each materialized model,${\sf MMQ}$first vectorizes the samples in the target dataset into probability vectors by directly applying this model, then utilizes Gaussian distribution to fit for each class of probability vectors, and finally uses separation degree on the Gaussian distributions to measure the target-related knowledge of the materialized model. Moreover, we propose an improved${\sf MMQ}$(${\sf I\text{-}MMQ}$), which significantly reduces the query time while retaining the query performance of${\sf MMQ}$. Extensive experiments on a range of practical model reuse workloads demonstrate the effectiveness and efficiency of${\sf MMQ}$.

Attributed community search aims to find the community with strong structure and attribute cohesiveness from attributed graphs. However, existing works suffer from two major limitations: (i) it is not easy to set the conditions on query attributes; (ii) the queries support only a single type of attributes. To make up for these deficiencies, in this paper, we study a novel attributed community search called vertex-centric attributed community (VAC) search. Given an attributed graph and a query vertex set, the VAC search returns the community which is densely connected (ensured by the k-truss model) and has the best attribute score. We show that the problem is NP-hard. To answer the VAC search, we develop both exact and approximate algorithms. Specifically, we develop two exact algorithms. One searches the community in a depth-first manner and the other is in a best-first manner. We also propose a set of heuristic strategies to prune the unqualified search space by exploiting the structure and attribute properties. In addition, to further improve the search efficiency, we propose a 2-approximation algorithm. Comprehensive experimental studies on various realworld attributed graphs demonstrate the effectiveness of the proposed model and the efficiency of the developed algorithms.

Community search enables personalized community discovery and has wide applications in large real-world graphs. While community search has been extensively studied for undirected graphs, the problem for directed graphs has received attention only recently. However, existing studies suffer from several drawbacks, e.g., the vertices with varied in-degrees and out-degrees cannot be included in a community at the same time. To address the limitations, in this paper, we systematically study the problem of community search over large directed graphs. We start by presenting a novel community model, called D-truss, based on two distinct types of directed triangles, i.e., flow triangle and cycle triangle. The D-truss model brings nice structural and computational properties and has many advantages in comparison with the existing models. With this new model, we then formulate the D-truss community search problem, which is proved to be NP-hard. In view of its hardness, we propose two efficient 2-approximation algorithms, named Global and Local, that run in polynomial time yet with quality guarantee. To further improve the efficiency of the algorithms, we devise an indexing method based on D-truss decomposition. Consequently, the D-truss community search can be solved upon the D-truss index without time-consuming accesses to the original graph. Experimental studies on real-world graphs with ground-truth communities validate the quality of the solutions we obtain and the efficiency of the proposed algorithms.