arXiv subjects

Soroush Omidvartehrani

Publications and source records attributed to Soroush Omidvartehrani.

3 recordsLinked to original sources

arXiv2026

LDI: Localized Data Imputation for Text-Rich Tables

Missing values are pervasive in real-world tabular data and can significantly impair downstream analysis. Imputing them is especially challenging in text-rich tables, where dependencies are implicit, complex, and dispersed across long textual fields. Recent work has explored using Large Language Models (LLMs) for data imputation, yet existing approaches typically process entire tables or loosely related contexts, which can compromise accuracy, scalability, and explainability. We introduce LDI, a novel framework that leverages LLMs through localized reasoning, selecting a compact, contextually relevant subset of attributes and tuples for each missing value. This targeted selection reduces noise, improves scalability, and provides transparent attribution by revealing the dependency relations that justify each selected attribute and the evidence behind each retrieved tuple. It makes clear not only which data influenced a prediction, but also why it was chosen. Through extensive experiments on real and synthetic datasets, we demonstrate that LDI consistently outperforms state-of-the-art imputation methods, achieving up to 8% higher accuracy with hosted LLMs and even greater gains with small local models. The improved interpretability and robustness also make LDI well-suited for high-stakes data management applications.

cs.DB↗

arXiv2026

EdgeLM: Edge Demonstrations for Language Models' Table Understanding

Large language models (LLMs) perform table-centric prediction through in-context learning, making demonstration selection critical to performance. Existing retrieval methods prioritize similarity to the query, but similar demonstrations often reinforce the model's likely prediction rather than reveal the distinctions needed for difficult decisions. We propose EdgeLM, a retrieval framework that instead selects edge evidence, demonstrations that are both relevant to the query and informative about the decision boundary. EdgeLM retrieves two complementary forms of edge evidence by selecting data edges, nearby examples with different ground-truth labels, and model edges, similar examples previously misclassified by the deployed model. EdgeLM requires neither model retraining nor task-specific engineering. Across five data wrangling tasks, fifteen datasets, and five open-weight and proprietary LLMs, EdgeLM consistently achieves the best or near-best performance in every setting, while ablations show that the two forms of edge evidence provide complementary benefits. Our code and datasets are publicly available at https://github.com/soroushomidvar/EdgeLM.

cs.CL↗

arXiv2025

GXJoin: Generalized Cell Transformations for Explainable Joinability

Describing real-world entities can vary across different sources, posing a challenge when integrating or exchanging data. We study the problem of joinability under syntactic transformations, where two columns are not equi-joinable but can become equi-joinable after some transformations. Discovering those transformations is a challenge because of the large space of possible candidates, which grows with the input length and the number of rows. Our focus is on the generality of transformations, aiming to make the relevant models applicable across various instances and domains. We explore a few generalization techniques, emphasizing those that yield transformations covering a larger number of rows and are often easier to explain. Through extensive evaluation on two real-world datasets and employing diverse metrics for measuring the coverage and simplicity of the transformations, our approach demonstrates superior performance over state-of-the-art approaches by generating fewer, simpler and hence more explainable transformations as well as improving the join performance.

cs.DB↗