Search

Pseudo relevance feedback (PRF) automatically performs query expansion based on top-retrieved documents to better represent the user’s information need so as to improve the search results. Previous PRF methods mainly select expansion terms with high occurrence frequency in top-retrieved documents or with high semantic similarity with the original query. However, existing PRF methods hardly try to understand the content of documents, which is very important in performing effective query expansion to reveal the user’s information need. In this paper, we propose a QA-based framework for PRF called QA4PRF to utilize contextual information in documents. In such a framework, we formulate PRF as a QA task, where the query and each top-retrieved document play the roles of question and context in the corresponding QA system, while the objective is to find some proper terms to expand the original query by utilizing contextual information, which are similar answers in QA task. Besides, an attention-based pointer network is built on understanding the content of top-retrieved documents and selecting the terms to represent the original query better. We also show that incorporating the traditional supervised learning methods, such as LambdaRank, to integrate PRF information will further improve the performance of QA4PRF. Extensive experiments on three real-world datasets demonstrate that QA4PRF significantly outperforms the state-of-the-art methods.

Besides well-known phase-locked loops, frequency-locked loops (FLLs) are applicable to grid synchronization for grid-tied power converters. The models, principles, and algorithm structures of existing notch filter-based single-phase FLLs are often described with different theories in previous studies. Consequently, a theoretically unified understanding and further advancements of such type of FLLs are restricted. This paper reinvestigates notch filter-based single-phase FLLs and develops a generalized design framework for such types of FLLs. The model derivations of the FLLs are made uniformly via the Popov's model reference adaptive control theory, in order to achieve the theoretically unified understanding. Three major contributions are received in this paper. First, a family of FLLs is derived, and existing seemingly different FLL designs are accordingly developed into a family of homologous FLLs. Second, the performance differences between the FLLs in the whole family are evaluated and compared quantitatively from the algorithm structure point of view, in terms of stability, convergence speed, and harmonic suppression capability. Finally, an enhanced single-phase FLL (named the 1φ-EFLL) is recommended by the family of the homologous FLLs. The designs and the performance evaluations are verified by both simulations and experiments.

Wind farms (WFs) controlled with conventional vector control (VC) algorithms cannot be directly integrated to the power grid through line commutated rectifier (LCR)-based high voltage direct current (HVDC) transmission due to the lack of voltage support at its sending-end bus. This paper proposes a novel coordinated control scheme for WFs with LCC-HVDC integration. The scheme comprises two key sub-control loops, referred to as the reactive power-based frequency (Q-f) control loop and the active power-based voltage (P-V) control loop, respectively. The Q-f control, applied to the voltage sources inverters in the WFs, maintains the system frequency and compensates the reactive power for the LCR of HVDC, whereas the P-V control, applied to the LCR, maintains the sending-end bus voltage and achieves the active power balance of the system. Phase-plane analysis and small-signal analysis are performed to evaluate the stability of the system and facilitate the controller parameter design. Simulations performed on PSCAD/EMTDC verify the proposed control scheme.

The article focuses on addressing biases in recommender systems caused by system-induced and user-induced factors. It introduces the concept of using a randomized dataset to mitigate system-induced biases and proposes a new theoretical framework to optimize the upper bound of an ideal objective function for debiasing. It also presents a novel method called "debiasing approximate upper bound (DUB)" and validates its effectiveness through extensive experiments on public and real product datasets.

The article's primary focus is on mitigating bias in recommendation systems through a novel framework based on user feature balancing. It explores strategies to handle biased user behavior data and introduces an upper bound learning objective that promotes unbiased recommendations. Topics covered include addressing selection bias, proposing strategies like clipping and sampling, and deploying an explicit model for handling latent confounders in recommendation systems.

The article primarily addresses the challenge of improving click-through rate (CTR) estimation in online personalization services by efficiently modeling long user behavior sequences. It introduces the User Behavior Retrieval (UBR) framework, emphasizing the use of learnable parametric retrieval functions to retrieve relevant behaviors for CTR estimation, leading to better performance and efficiency in real-world scenarios.

Learning to rank from logged user feedback, such as clicks or purchases, is a central component of many realworld information systems. Different from human-annotated relevance labels, the user feedback is always noisy and biased. Many existing learning to rank methods infer the underlying relevance of query--item pairs based on different assumptions of examination, and still optimize a relevance based objective. Such methods rely heavily on the correct estimation of examination, which is often difficult to achieve in practice. In this work, we propose a general framework U-rank+ for learning to rank with logged user feedback from the perspective of graph matching. We systematically analyze the biases in user feedback, including examination bias and selection bias. Then, we take both biases into consideration for unbiased utility estimation that directly based on user feedback, instead of relevance. In order to maximize the estimated utility in an efficient manner, we design two different solvers based on Sinkhorn and LambdaLoss for U-rank+. The former is based on a standard graph matching algorithm, and the latter is inspired by the traditional method of learning to rank. Both of the algorithms have good theoretical properties to optimize the unbiased utility objective while the latter is proved to be empirically more effective and efficient in practice. Our framework U-rank+ can deal with a general utility function and can be used in a widespread of applications including web search, recommendation, and online advertising. Semi-synthetic experiments on three benchmark learning to rank datasets demonstrate the effectiveness of U-rank+. Furthermore, our proposed framework has been deployed on two different scenarios of a mainstream App store, where the online A/B testing shows that U-rank+ achieves an average improvement of 19.2% on click-through rate and 20.8% improvement on conversion rate in recommendation scenario, and 5.12% on platform revenue in online advertising scenario over the production baselines. [ABSTRACT FROM AUTHOR]

With the rapid development of mobile app ecosystem, mobile apps have grown greatly popular. The explosive growth of apps makes it difficult for users to find apps that meet their interests. Therefore, it is necessary to recommend user with a personalized set of apps. However, one of the challenges is data sparsity, as users' historical behavior data are usually insufficient. In fact, user's behaviors from different domains in app store regarding the same apps are usually relevant. Therefore, we can alleviate the sparsity using complementary information from correlated domains. It is intuitive to model users' behaviors using graph, and graph neural networks have shown the great power for representation learning. In this article, we propose a novel model, Deep Multi-Graph Embedding (DMGE), to learn cross-domain app embedding. Specifically, we first construct a multi-graph based on users' behaviors from different domains, and then propose a multi-graph neural network to learn cross-domain app embedding. Particularly, we present an adaptivemethod to balance the weight of each domain and efficiently train the model. Finally, we achieve cross-domain app recommendation based on the learned app embedding. Extensive experiments on real-world datasets show that DMGE outperforms other state-of-art embedding methods. [ABSTRACT FROM AUTHOR]

User response prediction is a crucial component for personalized information retrieval and filtering scenarios, such as recommender system and web search. The data in user response prediction is mostly in a multi-field categorical format and transformed into sparse representations via one-hot encoding. Due to the sparsity problems in representation and optimization, most research focuses on feature engineering and shallow modeling. Recently, deep neural networks have attracted research attention on such a problem for their high capacity and end-to-end training scheme. In this article, we study user response prediction in the scenario of click prediction.We first analyze a coupled gradient issue in latent vector-based models and propose kernel product to learn field-aware feature interactions. Then, we discuss an insensitive gradient issue in DNN-based models and propose Product-based Neural Network, which adopts a feature extractor to explore feature interactions. Generalizing the kernel product to a net-in-net architecture,we further propose Productnetwork in Network (PIN), which can generalize previous models. Extensive experiments on four industrial datasets and one contest dataset demonstrate that ourmodels consistently outperform eight baselines on both area under curve and log loss. Besides, PIN makes great click-through rate improvement (relatively 34.67%) in online A/B test. [ABSTRACT FROM AUTHOR]

We consider the task model of periodic tasks running on a network of processor nodes connected by a bus based on the time-triggered protocol, an industry-standard bus protocol designed for safety-critical automotive and avionics distributed embedded systems, and present an integrated optimization framework that jointly considers one or more of the following attributes: task-to-processor allocation, task priority assignment, task period assignment and bus access configuration. We adopt a hierarchical optimization framework, where each possible task allocation and priority assignment is treated as one top-level coarse-grained state, which may contain many lower-level fine-grained states defined by different task period assignments and bus access configurations. Simulated annealing is used to explore the top-level states, which calls a geometric programming solver as a subroutine to explore the lower-level states contained within a given top-level state. Performance evaluation shows that our framework has good performance in terms of solution quality and scalability. [ABSTRACT FROM PUBLISHER]

Time-triggered protocol (TTP) is a time-division multiple access (TDMA)-based bus protocol designed for use in safety-critical avionics and automotive distributed embedded systems. Design space exploration (DSE) for TTP-based distributed embedded system involves searching through a vast design space of possible task-to-CPU mappings, task/message schedules and bus access configurations to achieve certain design objectives. In this paper, we present an efficient two-level hierarchical DSE framework for TTP-based distributed embedded systems, with the objective of minimizing the total bus utilization while meeting an end-to-end deadline constraint. Logic-based Benders decomposition (LBBD) is used to divide the problem into a master problem of mapping tasks to CPU nodes to minimize the total bus utilization, solved with a satisfiability modulo theories (SMT) solver, and a subproblem of finding a feasible solution of bus access configuration and task/message schedule under an end-to-end deadline constraint for a given task-to-CPU mapping, solved with a constraint programming (CP) solver. Performance evaluation results show that our approach is scalable to problems with realistic size. [ABSTRACT FROM PUBLISHER]