Researcher profile

Xiaolin Shi

Xiaolin Shi contributes to research discovery and scholarly infrastructure.

ResearcherAffiliation not importedOpen to collaborate

math.CO Applications cond-mat.dis-nn cond-mat.stat-mech Digital Libraries Machine Learning math.RT Social and Information Networks

Trust snapshot

Quick read

Trust 21 - Emerging

6works

0followers

8topics

4close collaborators

Actions

Decide how to stay connected

Follow researcher0

Inspect adjacent work, topics, institutions and collaborators without jumping out to a separate graph page.

Building this graph slice

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2020arXiv

Knowing your FATE: Friendship, Action and Temporal Explanations for User Engagement Prediction on Social Apps

With the rapid growth and prevalence of social network applications (Apps) in recent years, understanding user engagement has become increasingly important, to provide useful insights for future App design and development. While several promising neural modeling approaches were recently pioneered for accurate user engagement prediction, their black-box designs are unfortunately limited in model explainability. In this paper, we study a novel problem of explainable user engagement prediction for social network Apps. First, we propose a flexible definition of user engagement for various business scenarios, based on future metric expectations. Next, we design an end-to-end neural framework, FATE, which incorporates three key factors that we identify to influence user engagement, namely friendships, user actions, and temporal dynamics to achieve explainable engagement predictions. FATE is based on a tensor-based graph neural network (GNN), LSTM and a mixture attention mechanism, which allows for (a) predictive explanations based on learned weights across different feature categories, (b) reduced network complexity, and (c) improved performance in both prediction accuracy and training/inference time. We conduct extensive experiments on two large-scale datasets from Snapchat, where FATE outperforms state-of-the-art approaches by ${\approx}10\%$ error and ${\approx}20\%$ runtime reduction. We also evaluate explanations from FATE, showing strong quantitative and qualitative performance.

preprint2016arXiv

A Framework for Network AB Testing

A/B testing, also known as controlled experiment, bucket testing or splitting testing, has been widely used for evaluating a new feature, service or product in the data-driven decision processes of online websites. The goal of A/B testing is to estimate or test the difference between the treatment effects of the old and new variations. It is a well-studied two-sample comparison problem if each user's response is influenced by her treatment only. However, in many applications of A/B testing, especially those in HIVE of Yahoo and other social networks of Microsoft, Facebook, LinkedIn, Twitter and Google, users in the social networks influence their friends via underlying social interactions, and the conventional A/B testing methods fail to work. This paper considers the network A/B testing problem and provide a general framework consisting of five steps: data sampling, probabilistic model, parameter inference, computing average treatment effect and hypothesis test. The framework performs well for network A/B testing in simulation studies.

preprint2014arXiv

Bijective Projections on Parabolic Quotients of Affine Weyl Groups

Affine Weyl groups and their parabolic quotients are used extensively as indexing sets for objects in combinatorics, representation theory, algebraic geometry, and number theory. Moreover, in the classical Lie types we can conveniently realize the elements of these quotients via intuitive geometric and combinatorial models such as abaci, alcoves, coroot lattice points, core partitions, and bounded partitions. Berg, Jones, and Vazirani described a bijection between n-cores with first part equal to k and (n-1)-cores with first part less than or equal to k, and they interpret this bijection in terms of these other combinatorial models for the quotient of the affine symmetric group by the finite symmetric group. In this paper we discuss how to generalize the bijection of Berg-Jones-Vazirani to parabolic quotients of affine Weyl groups in type C. We develop techniques using the associated affine hyperplane arrangement to interpret this bijection geometrically as a projection of alcoves onto the hyperplane containing their coroot lattice points. We are thereby able to analyze this bijective projection in the language of various additional combinatorial models developed by Hanusa and Jones, such as abaci, core partitions, and canonical reduced expressions in the Coxeter group.

preprint2012arXiv

On Orbits of Order Ideals of Minuscule Posets

An action on order ideals of posets considered by Fon-Der-Flaass is analyzed in the case of posets arising from minuscule representations of complex simple Lie algebras. For these minuscule posets, it is shown that the Fon-Der-Flaass action exhibits the cyclic sieving phenomenon, as defined by Reiner, Stanton, and White. A uniform proof is given by investigation of a bijection due to Stembridge between order ideals of minuscule posets and fully commutative Weyl group elements. This bijection is proven to be equivariant with respect to a conjugate of the Fon-Der-Flaass action and an arbitrary Coxeter element. If $P$ is a minuscule poset, it is shown that the Fon-Der-Flaass action on order ideals of the Cartesian product $P \times [2]$ also exhibits the cyclic sieving phenomenon, only the proof is by appeal to the classification of minuscule posets and is not uniform.

preprint2010arXiv

Citing for High Impact

The question of citation behavior has always intrigued scientists from various disciplines. While general citation patterns have been widely studied in the literature we develop the notion of citation projection graphs by investigating the citations among the publications that a given paper cites. We investigate how patterns of citations vary between various scientific disciplines and how such patterns reflect the scientific impact of the paper. We find that idiosyncratic citation patterns are characteristic for low impact papers; while narrow, discipline-focused citation patterns are common for medium impact papers. Our results show that crossing-community, or bridging citation patters are high risk and high reward since such patterns are characteristic for both low and high impact papers. Last, we observe that recently citation networks are trending toward more bridging and interdisciplinary forms.

preprint2006arXiv

Networks of strong ties

Social networks transmitting covert or sensitive information cannot use all ties for this purpose. Rather, they can only use a subset of ties that are strong enough to be ``trusted''. In this paper we consider transitivity as evidence of strong ties, requiring that each tie can only be used if the individuals on either end also share at least one other contact in common. We examine the effect of removing all non-transitive ties in two real social network data sets. We observe that although some individuals become disconnected, a giant connected component remains, with an average shortest path only slightly longer than that of the original network. We also evaluate the cost of forming transitive ties by deriving the conditions for the emergence and the size of the giant component in a random graph composed entirely of closed triads and the equivalent Erdos-Renyi random graph.