S. Muthukrishnan

Yikun Xian, Zuohui Fu, Handong Zhao et al.

Recent research explores incorporating knowledge graphs (KG) into e-commerce recommender systems, not only to achieve better recommendation performance, but more importantly to generate explanations of why particular decisions are made. This can be achieved by explicit KG reasoning, where a model starts from a user node, sequentially determines the next step, and walks towards an item node of potential interest to the user. However, this is challenging due to the huge search space, unknown destination, and sparse signals over the KG, so informative and effective guidance is needed to achieve a satisfactory recommendation quality. To this end, we propose a CoArse-to-FinE neural symbolic reasoning approach (CAFE). It first generates user profiles as coarse sketches of user behaviors, which subsequently guide a path-finding process to derive reasoning paths for recommendations as fine-grained predictions. User profiles can capture prominent user behaviors from the history, and provide valuable signals about which kinds of path patterns are more likely to lead to potential items of interest for the user. To better exploit the user profiles, an improved path-finding algorithm called Profile-guided Path Reasoning (PPR) is also developed, which leverages an inventory of neural symbolic reasoning modules to effectively and efficiently find a batch of paths over a large-scale KG. We extensively experiment on four real-world benchmarks and observe substantial gains in the recommendation performance compared with state-of-the-art methods.

Yikun Xian, Zuohui Fu, Qiaoying Huang et al.

Recent work on recommender systems has considered external knowledge graphs as valuable sources of information, not only to produce better recommendations but also to provide explanations of why the recommended items were chosen. Pure rule-based symbolic methods provide a transparent reasoning process over knowledge graph but lack generalization ability to unseen examples, while deep learning models enhance powerful feature representation ability but are hard to interpret. Moreover, direct reasoning over large-scale knowledge graph can be costly due to the huge search space of pathfinding. We approach the problem through a novel coarse-to-fine neural symbolic reasoning method called NSER. It first generates a coarse-grained explanation to capture abstract user behavioral pattern, followed by a fined-grained explanation accompanying with explicit reasoning paths and recommendations inferred from knowledge graph. We extensively experiment on four real-world datasets and observe substantial gains of recommendation performance compared with state-of-the-art methods as well as more diversified explanations in different granularity.

Yikun Xian, Zuohui Fu, S. Muthukrishnan et al.

Recent advances in personalized recommendation have sparked great interest in the exploitation of rich structured information provided by knowledge graphs. Unlike most existing approaches that only focus on leveraging knowledge graphs for more accurate recommendation, we perform explicit reasoning with knowledge for decision making so that the recommendations are generated and supported by an interpretable causal inference procedure. To this end, we propose a method called Policy-Guided Path Reasoning (PGPR), which couples recommendation and interpretability by providing actual paths in a knowledge graph. Our contributions include four aspects. We first highlight the significance of incorporating knowledge graphs into recommendation to formally define and interpret the reasoning process. Second, we propose a reinforcement learning (RL) approach featuring an innovative soft reward strategy, user-conditional action pruning and a multi-hop scoring function. Third, we design a policy-guided graph search algorithm to efficiently and effectively sample reasoning paths for recommendation. Finally, we extensively evaluate our method on several large-scale real-world benchmark datasets, obtaining favorable results compared with state-of-the-art methods.

Branislav Kveton, Saied Mahdian, S. Muthukrishnan et al.

A popular approach to selling online advertising is by a waterfall, where a publisher makes sequential price offers to ad networks for an inventory, and chooses the winner in that order. The publisher picks the order and prices to maximize her revenue. A traditional solution is to learn the demand model and then subsequently solve the optimization problem for the given demand model. This will incur a linear regret. We design an online learning algorithm for solving this problem, which interleaves learning and optimization, and prove that this algorithm has sublinear regret. We evaluate the algorithm on both synthetic and real-world data, and show that it quickly learns high quality pricing strategies. This is the first principled study of learning a waterfall design online by sequential experimentation.

Shuai Li, Yasin Abbasi-Yadkori, Branislav Kveton et al.

Many web systems rank and present a list of items to users, from recommender systems to search and advertising. An important problem in practice is to evaluate new ranking policies offline and optimize them before they are deployed. We address this problem by proposing evaluation algorithms for estimating the expected number of clicks on ranked lists from historical logged data. The existing algorithms are not guaranteed to be statistically efficient in our problem because the number of recommended lists can grow exponentially with their length. To overcome this challenge, we use models of user interaction with the list of items, the so-called click models, to construct estimators that learn statistically efficiently. We analyze our estimators and prove that they are more efficient than the estimators that do not use the structure of the click model, under the assumption that the click model holds. We evaluate our estimators in a series of experiments on a real-world dataset and show that they consistently outperform prior estimators.

Branislav Kveton, Csaba Szepesvari, Anup Rao et al.

Many problems in computer vision and recommender systems involve low-rank matrices. In this work, we study the problem of finding the maximum entry of a stochastic low-rank matrix from sequential observations. At each step, a learning agent chooses pairs of row and column arms, and receives the noisy product of their latent values as a reward. The main challenge is that the latent values are unobserved. We identify a class of non-negative matrices whose maximum entry can be found statistically efficiently and propose an algorithm for finding them, which we call LowRankElim. We derive a $\DeclareMathOperator{\poly}{poly} O((K + L) \poly(d) Δ^{-1} \log n)$ upper bound on its $n$-step regret, where $K$ is the number of rows, $L$ is the number of columns, $d$ is the rank of the matrix, and $Δ$ is the minimum gap. The bound depends on other problem-specific constants that clearly do not depend $K L$. To the best of our knowledge, this is the first such result in the literature.

Branislav Kveton, Hung Bui, Mohammad Ghavamzadeh et al.

Structured high-cardinality data arises in many domains, and poses a major challenge for both modeling and inference. Graphical models are a popular approach to modeling structured data but they are unsuitable for high-cardinality variables. The count-min (CM) sketch is a popular approach to estimating probabilities in high-cardinality data but it does not scale well beyond a few variables. In this work, we bring together the ideas of graphical models and count sketches; and propose and analyze several approaches to estimating probabilities in structured high-cardinality streams of data. The key idea of our approximations is to use the structure of a graphical model and approximately estimate its factors by "sketches", which hash high-cardinality variables using random projections. Our approximations are computationally efficient and their space complexity is independent of the cardinality of variables. Our error bounds are multiplicative and significantly improve upon those of the CM sketch, a state-of-the-art approach to estimating probabilities in streams. We evaluate our approximations on synthetic and real-world problems, and report an order of magnitude improvements over the CM sketch.

Stratis Ioannidis, S. Muthukrishnan, Jinyun Yan

In many cases, recommendations are consumed by groups of users rather than individuals. In this paper, we present a system which recommends social events to groups. The system helps groups to organize a joint activity and collectively select which activity to perform among several possible options. We also facilitate the consensus making, following the principle of group consensus decision making. Our system allows users to asynchronously vote, add and comment on alternatives. We observe social influence within groups through post-recommendation feedback during the group decision making process. We propose a decision cascading model and estimate such social influence, which can be used to improve the performance of group recommendation. We conduct experiments to measure the prediction performance of our model. The result shows that the model achieves better results than that of independent decision making model.

Santiago Balseiro, Jon Feldman, Vahab Mirrokni et al.

In light of the growing market of Ad Exchanges for the real-time sale of advertising slots, publishers face new challenges in choosing between the allocation of contract-based reservation ads and spot market ads. In this setting, the publisher should take into account the tradeoff between short-term revenue from an Ad Exchange and quality of allocating reservation ads. In this paper, we formalize this combined optimization problem as a stochastic control problem and derive an efficient policy for online ad allocation in settings with general joint distribution over placement quality and exchange bids. We prove asymptotic optimality of this policy in terms of any trade-off between quality of delivered reservation ads and revenue from the exchange, and provide a rigorous bound for its convergence rate to the optimal policy. We also give experimental results on data derived from real publisher inventory, showing that our policy can achieve any pareto-optimal point on the quality vs. revenue curve. Finally, we study a parametric training-based algorithm in which instead of learning the dual variables from a sample data (as is done in non-parametric training-based algorithms), we learn the parameters of the distribution and construct those dual variables from the learned parameter values. We compare parametric and non-parametric ways to estimate from data both analytically and experimentally in the special case without the ad exchange, and show that though both methods converge to the optimal policy as the sample size grows, our parametric method converges faster, and thus performs better on smaller samples.