arXiv subjects

Alex Thomo

Publications and source records attributed to Alex Thomo.

15 recordsLinked to original sources

arXiv2026

LLMs for Social Network Modeling: From Network Generation to Dynamic Processes

Large language models (LLMs) are rapidly emerging as a new paradigm for modeling social networks by representing users and their relationships and interactions through natural language. Unlike classical network models or deep learning approaches, LLMs can simulate context-aware social behavior and language-driven interactions, enabling more realistic modeling of network formation and dynamic social processes. However, existing studies are scattered across different research communities and lack a unified perspective. This survey presents the first comprehensive review of LLMs for social network modeling by organizing the literature into two broad categories: network generative models and dynamic process models. Network generative models are further classified into selection-based and interaction-based approaches, while dynamic process models are categorized into opinion dynamics, information diffusion, and rumor propagation, each with their underlying modeling mechanisms. LLMs enable rich textual social interactions and decision-making, but they also exhibit many limitations, including inherent social biases and prompt sensitivity. We outline these open research challenges and discuss future directions in LLM-based social network modeling.

cs.SI↗

arXiv2026

The Reasoning Bottleneck in Graph-RAG: Structured Prompting and Context Compression for Multi-Hop QA

Graph-RAG systems achieve strong multi-hop question answering by indexing documents into knowledge graphs, but strong retrieval does not guarantee strong answers. Evaluating KET-RAG, a leading Graph-RAG system, on three multi-hop QA benchmarks (HotpotQA, MuSiQue, 2WikiMultiHopQA), we find that 77% to 91% of questions have the gold answer in the retrieved context, yet accuracy is only 35% to 78%, and 73% to 84% of errors are reasoning failures. We propose two augmentations: (i) SPARQL chain-of-thought prompting, which decomposes questions into triple-pattern queries aligned with the entity-relationship context, and (ii) graph-walk compression, which compresses the context by ~60% via knowledge-graph traversal with no LLM calls. SPARQL CoT improves accuracy by +2 to +14 pp; graph-walk compression adds +6 pp on average when paired with structured prompting on smaller models. Surprisingly, we show that, with question-type routing, a fully augmented budget open-weight Llama-8B model matches or exceeds the unaugmented Llama-70B baseline on all three benchmarks at ~12x lower cost. A replication on LightRAG confirms that our augmentations transfer across Graph-RAG systems.

cs.IR↗

arXiv2026

Gender and Race Bias in Consumer Product Recommendations by Large Language Models

Large Language Models are increasingly employed in generating consumer product recommendations, yet their potential for embedding and amplifying gender and race biases remains underexplored. This paper serves as one of the first attempts to examine these biases within LLM-generated recommendations. We leverage prompt engineering to elicit product suggestions from LLMs for various race and gender groups and employ three analytical methods-Marked Words, Support Vector Machines, and Jensen-Shannon Divergence-to identify and quantify biases. Our findings reveal significant disparities in the recommendations for demographic groups, underscoring the need for more equitable LLM recommendation systems.

cs.CL↗

arXiv2022

Computing (1+epsilon)-Approximate Degeneracy in Sublinear Time

The problem of finding the degeneracy of a graph is a subproblem of the k-core decomposition problem. In this paper, we present a (1 + epsilon)-approximate solution to the degeneracy problem which runs in O(n log n) time, sublinear in the input size for dense graphs, by sampling a small number of neighbors adjacent to high degree nodes. Our algorithm can also be extended to an O(n log n) time solution to the k-core decomposition problem. This improves upon the method by Bhattacharya et al., which implies a (4 + epsilon)-approximate ~O(n) solution to the degeneracy problem, and our techniques are similar to other sketching methods which use sublinear space for k-core and degeneracy. We prove theoretical guarantees of our algorithm and provide optimizations, which improve the running time of our algorithm in practice. Experiments on massive real-world web graphs show that our algorithm performs significantly faster than previous methods for computing degeneracy, including the 2022 exact degeneracy algorithm by Li et al.

cs.DS↗

arXiv2021

Multi-Stage Graph Peeling Algorithm for Probabilistic Core Decomposition

Mining dense subgraphs where vertices connect closely with each other is a common task when analyzing graphs. A very popular notion in subgraph analysis is core decomposition. Recently, Esfahani et al. presented a probabilistic core decomposition algorithm based on graph peeling and Central Limit Theorem (CLT) that is capable of handling very large graphs. Their proposed peeling algorithm (PA) starts from the lowest degree vertices and recursively deletes these vertices, assigning core numbers, and updating the degree of neighbour vertices until it reached the maximum core. However, in many applications, particularly in biology, more valuable information can be obtained from dense sub-communities and we are not interested in small cores where vertices do not interact much with others. To make the previous PA focus more on dense subgraphs, we propose a multi-stage graph peeling algorithm (M-PA) that has a two-stage data screening procedure added before the previous PA. After removing vertices from the graph based on the user-defined thresholds, we can reduce the graph complexity largely and without affecting the vertices in subgraphs that we are interested in. We show that M-PA is more efficient than the previous PA and with the properly set filtering threshold, can produce very similar if not identical dense subgraphs to the previous PA (in terms of graph density and clustering coefficient).

stat.ML↗

arXiv2020

Utility-Based Graph Summarization: New and Improved

A fundamental challenge in graph mining is the ever-increasing size of datasets. Graph summarization aims to find a compact representation resulting in faster algorithms and reduced storage needs. The flip side of graph summarization is the loss of utility which diminishes its usability. The key questions we address in this paper are: (1)How to summarize a graph without any loss of utility? (2)How to summarize a graph with some loss of utility but above a user-specified threshold? (3)How to query graph summaries without graph reconstruction?} We also aim at making graph summarization available for the masses by efficiently handling web-scale graphs using only a consumer-grade machine. Previous works suffer from conceptual limitations and lack of scalability. In this work, we make three key contributions. First, we present a utility-driven graph summarization method, based on a clique and independent set decomposition, that produces significant compression with zero loss of utility. The compression provided is significantly better than state-of-the-art in lossless graph summarization, while the runtime is two orders of magnitude lower. Second, we present a highly scalable algorithm for the lossy case, which foregoes the expensive iterative process that hampers previous work. Our algorithm achieves this by combining a memory reduction technique and a novel binary-search approach. In contrast to the competition, we are able to handle web-scale graphs in a single machine without a performance impediment as the utility threshold (and size of summary) decreases. Third, we show that our graph summaries can be used as-is to answer several important classes of queries, such as triangle enumeration, Pagerank, and shortest paths. This is in contrast to other works that incrementally reconstruct the original graph for answering queries, thus incurring additional time costs.

cs.DS↗

arXiv2020

Nucleus Decomposition in Probabilistic Graphs: Hardness and Algorithms

Finding dense components in graphs is of great importance in analyzing the structure of networks. Popular and computationally feasible frameworks for discovering dense subgraphs are core and truss decompositions. Recently, Sariyuce et al. introduced nucleus decomposition, a generalization which uses higher-order structures and can reveal interesting subgraphs that can be missed by core and truss decompositions. In this paper, we present nucleus decomposition in probabilistic graphs. We study the most interesting case of nucleus decomposition, k-(3,4)-nucleus, which asks for maximal subgraphs where each triangle is contained in k 4-cliques. The major questions we address are: How to define meaningfully nucleus decomposition in probabilistic graphs? How hard is computing nucleus decomposition in probabilistic graphs? Can we devise efficient algorithms for exact or approximate nucleus decomposition in large graphs? We present three natural definitions of nucleus decomposition in probabilistic graphs: local, global, and weakly-global. We show that the local version is in PTIME, whereas global and weakly-global are #P-hard and NP-hard, respectively. We present an efficient and exact dynamic programming approach for the local case and furthermore, present statistical approximations that can scale to large datasets without much loss of accuracy. For global and weakly-global decompositions, we complement our intractability results by proposing efficient algorithms that give approximate solutions based on search space pruning and Monte-Carlo sampling. Our extensive experimental results show the scalability and efficiency of our algorithms. Compared to probabilistic core and truss decompositions, nucleus decomposition significantly outperforms in terms of density and clustering metrics.

cs.SI↗

arXiv2020

Provenance for Regular Path Queries

Regular path queries (RPQs) the ubiquitous mechanism for querying data graphs of partially known structure. RPQs are in essence regular expressions over the edge symbols. The answer to an RPQ on a given graph (database) is the set of pairs of objects, which are connected by paths spelling words in the language of the regular path query. Often the database edges come with a weights assoaciated to them. Such weights can distances, levels of discomfort, multiplicities, etc. We model weights using semiring frameworks.

cs.FL↗

arXiv2018

Reverse Prevention Sampling for Misinformation Mitigation in Social Networks

In this work, we consider misinformation propagating through a social network and study the problem of its prevention. In this problem, a "bad" campaign starts propagating from a set of seed nodes in the network and we use the notion of a limiting (or "good") campaign to counteract the effect of misinformation. The goal is to identify a set of $k$ users that need to be convinced to adopt the limiting campaign so as to minimize the number of people that adopt the "bad" campaign at the end of both propagation processes. This work presents \emph{RPS} (Reverse Prevention Sampling), an algorithm that provides a scalable solution to the misinformation mitigation problem. Our theoretical analysis shows that \emph{RPS} runs in $O((k + l)(n + m)(\frac{1}{1 - \gamma}) \log n / \epsilon^2 )$ expected time and returns a $(1 - 1/e - \epsilon)$-approximate solution with at least $1 - n^{-l}$ probability (where $\gamma$ is a typically small network parameter and $l$ is a confidence parameter). The time complexity of \emph{RPS} substantially improves upon the previously best-known algorithms that run in time $\Omega(m n k \cdot POLY(\epsilon^{-1}))$. We experimentally evaluate \emph{RPS} on large datasets and show that it outperforms the state-of-the-art solution by several orders of magnitude in terms of running time. This demonstrates that misinformation mitigation can be made practical while still offering strong theoretical guarantees.

cs.SI↗

arXiv2017

Computation of K-Core Decomposition on Giraph

Graphs are an essential data structure that can represent the structure of social networks. Many online companies, in order to provide intelligent and personalized services for their users, aim to comprehensively analyze a significant amount of graph data with different features. One example is k-core decomposition which captures the degree of connectedness in social graphs. The main purpose of this report is to explore a distributed algorithm for k-core decomposition on Apache Giraph. Namely, we would like to determine whether a cluster-based, Giraph implementation of k-core decomposition that we provide is more efficient than a single-machine, disk-based implementation on GraphChi for large networks. In this report, we describe (a) the programming model of Giraph and GraphChi, (b) the specific implementation of k-core decomposition with Giraph, and (c) the result comparison between Giraph and GraphChi. By analyzing the results, we conclude that Giraph is faster than GraphChi when dealing with large data. However, since worker nodes need time to communicate with each other, Giraph is not very efficient for small data.

cs.DS↗

arXiv2014

Clearing Contamination in Large Networks

In this work, we study the problem of clearing contamination spreading through a large network where we model the problem as a graph searching game. The problem can be summarized as constructing a search strategy that will leave the graph clear of any contamination at the end of the searching process in as few steps as possible. We show that this problem is NP-hard even on directed acyclic graphs and provide an efficient approximation algorithm. We experimentally observe the performance of our approximation algorithm in relation to the lower bound on several large online networks including Slashdot, Epinions and Twitter. The experiments reveal that in most cases our algorithm performs near optimally.

cs.SI↗

arXiv2014

Buyer to Seller Recommendation under Constraints

The majority of recommender systems are designed to recommend items (such as movies and products) to users. We focus on the problem of recommending buyers to sellers which comes with new challenges: (1) constraints on the number of recommendations buyers are part of before they become overwhelmed, (2) constraints on the number of recommendations sellers receive within their budget, and (3) constraints on the set of buyers that sellers want to receive (e.g., no more than two people from the same household). We propose the following critical problems of recommending buyers to sellers: Constrained Recommendation (C-REC) capturing the first two challenges, and Conflict-Aware Constrained Recommendation (CAC-REC) capturing all three challenges at the same time. We show that C-REC can be modeled using linear programming and can be efficiently solved using modern solvers. On the other hand, we show that CAC-REC is NP-hard. We propose two approximate algorithms to solve CAC-REC and show that they achieve close to optimal solutions via comprehensive experiments using real-world datasets.

cs.SI↗

arXiv2014

Three-Way Joins on MapReduce: An Experimental Study

We study three-way joins on MapReduce. Joins are very useful in a multitude of applications from data integration and traversing social networks, to mining graphs and automata-based constructions. However, joins are expensive, even for moderate data sets; we need efficient algorithms to perform distributed computation of joins using clusters of many machines. MapReduce has become an increasingly popular distributed computing system and programming paradigm. We consider a state-of-the-art MapReduce multi-way join algorithm by Afrati and Ullman and show when it is appropriate for use on very large data sets. By providing a detailed experimental study, we demonstrate that this algorithm scales much better than what is suggested by the original paper. However, if the join result needs to be summarized or aggregated, as opposed to being only enumerated, then the aggregation step can be integrated into a cascade of two-way joins, making it more efficient than the other algorithm, and thus becomes the preferred solution.

cs.DB↗

arXiv2012

Computing optimal k-regret minimizing sets with top-k depth contours

Regret minimizing sets are a very recent approach to representing a dataset D with a small subset S of representative tuples. The set S is chosen such that executing any top-1 query on S rather than D is minimally perceptible to any user. To discover an optimal regret minimizing set of a predetermined cardinality is conjectured to be a hard problem. In this paper, we generalize the problem to that of finding an optimal k$regret minimizing set, wherein the difference is computed over top-k queries, rather than top-1 queries. We adapt known geometric ideas of top-k depth contours and the reverse top-k problem. We show that the depth contours themselves offer a means of comparing the optimality of regret minimizing sets using L2 distance. We design an O(cn^2) plane sweep algorithm for two dimensions to compute an optimal regret minimizing set of cardinality c. For higher dimensions, we introduce a greedy algorithm that progresses towards increasingly optimal solutions by exploiting the transitivity of L2 distance.

cs.DB↗

arXiv2012

Indexing Reverse Top-k Queries

We consider the recently introduced monochromatic reverse top-k queries which ask for, given a new tuple q and a dataset D, all possible top-k queries on D union {q} for which q is in the result. Towards this problem, we focus on designing indexes in two dimensions for repeated (or batch) querying, a novel but practical consideration. We present the insight that by representing the dataset as an arrangement of lines, a critical k-polygon can be identified and used exclusively to respond to reverse top-k queries. We construct an index based on this observation which has guaranteed worst-case query cost that is logarithmic in the size of the k-polygon. We implement our work and compare it to related approaches, demonstrating that our index is fast in practice. Furthermore, we demonstrate through our experiments that a k-polygon is comprised of a small proportion of the original data, so our index structure consumes little disk space.

cs.DB↗