publications

2026

1. SSRN

   Description Logic Axiom Construction: A Narrative Review of Symbolic, Triple-Based, Lexical, and Large Language Model Approaches

   Aryan Dalal, Soheil Abadifard, Nirmal Gelal, Pravesh Niroula, Kathleen Jagodnik, and Hande McGinty

   SSRN Electronic Journal, Oct 2026

   Preprint

   Abs DOI Bib

   Constructing expressive axioms in the Web Ontology Language and description logics, such as subclass, disjointness, domain, range, and general concept inclusion axioms, remains a bottleneck in ontology learning, because manual authoring does not scale to large knowledge graphs. Recent surveys have covered parts of this field, but none that we identified places lexical methods, symbolic and neural-symbolic description logic learners, triple-based inductive and logical methods, and large language model methods side by side on shared dimensions. This narrative survey unifies these four families under a single taxonomy of axiom construction over Resource Description Framework data and related inputs, including triples, text, labeled examples, ontology signatures, and prompts. We compare the families by input requirements, axiom coverage, controllability, scalability, and dependence on triples. We identify a recurring architectural pattern in the language model literature, in which a generative model proposes candidate axioms and a verification layer based on retrieval, schema repair, entailment, methodology audit, or description logic reasoning constrains them. Because the studies use different benchmarks and metrics, we do not treat their results as directly comparable, and we discuss a hybrid pipeline argument in which lexical and language model proposers, triple-based scorers, and description logic learners and reasoners each carry the operational load they are best suited for. The principal open problem is evaluation, especially benchmark leakage in the language model era.

   @article{dalal2026dlaxioms,
     title = {Description Logic Axiom Construction: A Narrative Review of Symbolic, Triple-Based, Lexical, and Large Language Model Approaches},
     author = {Dalal, Aryan and Abadifard, Soheil and Gelal, Nirmal and Niroula, Pravesh and Jagodnik, Kathleen and McGinty, Hande},
     journal = {SSRN Electronic Journal},
     note = {Preprint},
     year = {2026},
     month = oct,
     publisher = {Elsevier BV},
     doi = {10.2139/ssrn.7578678},
     url = {https://doi.org/10.2139/ssrn.7578678},
   }

2. SSRN

   Infrastructure-Aware Urban Flood Susceptibility from Multi-Event High-Water Marks: A Multi-Model Comparison

   Alireza Monavarian, Soheil Abadifard, William J. Heatherman, Hande McGinty, and Vaishali Sharda

   SSRN Electronic Journal, Oct 2026

   Preprint

   Abs DOI Bib Code

   Urban flooding is driven by pluvial and infrastructure-mediated processes, yet flood susceptibility models typically rely on terrain-centric predictors and on flood inventories or remotely sensed extents. This study evaluates two underused inputs, high-water marks surveyed by municipal crews as presence observations and a city stormwater geodatabase as a source of infrastructure predictors, such as outfall proximity. Using 277 presence points from 11 flood events (2007–2024) in Manhattan, Kansas, and 17 predictors at 10 m resolution, we fitted a maximum-entropy model, compared it with five supervised classifiers, and assessed generalization by spatial-block and leave-one-event-out cross-validation. Distance to the nearest outfall was the strongest single predictor (area under the receiver-operating-characteristic curve, AUC = 0.915). Spatial-block and event cross-validation gave AUCs of 0.935 and 0.952, respectively, against 0.983 for a random split, and no model differed significantly from another after Holm correction. After controlling for road access and four terrain and land-cover confounders, outfall proximity still added 0.025 AUC. The regulatory 1% floodplain captured 40% of held-out marks while covering 17% of the study area, whereas the model captured 95% within 7%. Refitted on marks outside the floodplain, distance to outfall alone still reached AUC = 0.920, so the association is not limited to riverine flooding. An exploratory socioeconomic overlay found no significant association after Holm correction. Proximity to stormwater outfalls was the strongest available correlate of observed flooding. Because these two inputs are municipal records and the code is openly available, the workflow can be adapted to other cities that hold them.

   @article{monavarian2026flood,
     title = {Infrastructure-Aware Urban Flood Susceptibility from Multi-Event High-Water Marks: A Multi-Model Comparison},
     author = {Monavarian, Alireza and Abadifard, Soheil and Heatherman, William J. and McGinty, Hande and Sharda, Vaishali},
     journal = {SSRN Electronic Journal},
     note = {Preprint},
     year = {2026},
     month = oct,
     publisher = {Elsevier BV},
     doi = {10.2139/ssrn.7583892},
     url = {https://doi.org/10.2139/ssrn.7583892},
   }

3. SSRN

   From Field Data to Global Food Systems Intelligence: A Semantic Graph Framework for Sustainable Wheat Production

   Nirmal Gelal, Aastha Gautam, Soheil Abadifard, Nicolas Giordano, Moumita Sen Sarma, Sanaz Saki Norouzi, Claudio Dias da Silva Jr, Jean Ribert Francois, Kathleen M. Jagodnik, Katherine Nelson, Terry Griffin, Xiaomao Lin, Stacy Hutchinson, Stephen Welch, Kelsey Andersen Onofre, Romulo Lollato, Pascal Hitzler, and Hande McGinty

   SSRN Electronic Journal, Sep 2026

   Preprint. An earlier version is on arXiv.

   Abs DOI arXiv Bib Code Website

   In response to the growing need for structured, interoperable agricultural data, this paper presents the Sustainable Wheat Production Datahub, a modular, graph-based framework that brings diverse wheat production datasets together into a single, queryable store. Using the Knowledge Acquisition and Representation Methodology (KNARM), with domain experts in the loop, we developed ontologies for nutrient management and disease management, two of the areas that most affect wheat yield and its sustainability, covering practices such as nitrogen fertilization and fungicide-based disease control. The two ontologies are federated, meaning they are maintained as separate but connected modules, joined by a bridging layer of cross-domain links, and together they form the schema of a knowledge graph (KG). We construct this KG by populating the ontologies with data from diverse sources, including field trials, expert knowledge, and environmental descriptors. We validate the ontologies, showing the KG accurately answers practical agronomic questions expressed in SPARQL. We demonstrate that the KG derives facts entailed by the ontologies beyond those explicitly stored, and that a single query can draw across independently sourced datasets. The framework also has practical implications for wheat research and extension programs, since it makes data from different sources easier to find, combine, and reuse. Designed to expand toward the full wheat lifecycle, from farm to table, this work establishes the foundation for a scalable, semantically rich global food systems datahub.

   @article{gelal2026wheat,
     title = {From Field Data to Global Food Systems Intelligence: A Semantic Graph Framework for Sustainable Wheat Production},
     author = {Gelal, Nirmal and Gautam, Aastha and Abadifard, Soheil and Giordano, Nicolas and Sarma, Moumita Sen and Norouzi, Sanaz Saki and {Dias da Silva Jr}, Claudio and Francois, Jean Ribert and Jagodnik, Kathleen M. and Nelson, Katherine and Griffin, Terry and Lin, Xiaomao and Hutchinson, Stacy and Welch, Stephen and {Andersen Onofre}, Kelsey and Lollato, Romulo and Hitzler, Pascal and McGinty, Hande},
     journal = {SSRN Electronic Journal},
     note = {Preprint. An earlier version is on arXiv.},
     year = {2026},
     month = sep,
     publisher = {Elsevier BV},
     doi = {10.2139/ssrn.7532165},
     url = {https://doi.org/10.2139/ssrn.7532165},
   }

4. SSRN

   Bioinformatics Pipeline for Analyzing Space-Flown Mammary Gland

   Soheil Abadifard, Maria Baloch, Joseph Guarnieri, Begum Mathyk, and Hande McGinty

   SSRN Electronic Journal, May 2026

   Preprint. Presented as posters at US2TS 2025 and ASGSR 2025.

   Abs DOI Bib Code

   This study investigates how age affects the biological response of mammary gland tissue to spaceflight, focusing on potential risks for breast and ovarian cancers. We introduced a novel bioinformatics pipeline to analyze tissues from young and old mice which were sent to space. Through our analysis, we found that older mice had a significantly stronger molecular response to space flight elements. Key biological processes, such as signal transduction and gene regulation, were observed to be more active in the older group. The molecular changes in older mice showed a stronger association with genes linked to cancer, particularly ovarian and breast cancer. Our analysis points to conclusions that show how age is a critical factor in how the body responds to spaceflight. This study points to specific biological pathways that could be targeted for future cancer risk assessment and prevention for astronauts as well as assessing reproductive system risks for space missions.

   @article{abadifard2026mammary,
     title = {Bioinformatics Pipeline for Analyzing Space-Flown Mammary Gland},
     author = {Abadifard, Soheil and Baloch, Maria and Guarnieri, Joseph and Mathyk, Begum and McGinty, Hande},
     journal = {SSRN Electronic Journal},
     note = {Preprint. Presented as posters at US2TS 2025 and ASGSR 2025.},
     year = {2026},
     month = may,
     publisher = {Elsevier BV},
     doi = {10.2139/ssrn.6663438},
     url = {https://doi.org/10.2139/ssrn.6663438},
   }

5. Land

   Machine Learning-Based Bioclimatic Suitability Modeling for Maize Cultivation Under Future Projections

   Alireza Monavarian, Soheil Abadifard, Hande K. McGinty, and Vaishali Sharda

   Land, Apr 2026

   Abs DOI Bib

   Climate-driven heat and water stress are increasingly compromising rainfed maize yields in transition zones, with significant implications for global food security. While continental-scale models of crop suitability exist, they often fail to capture the high-resolution heterogeneity of agricultural landscapes or distinguish between irrigated and rainfed systems in semi-arid regions. This study models the current and future suitability of rainfed maize in Kansas, USA, using a Maximum Entropy (MaxEnt) approach. To accurately isolate biophysical constraints, we employed a novel data-filtering workflow using the USDA Cropland Data Layer (CDL) and Landsat-based Annual Irrigated Datasets (LANID) to train the model exclusively on rainfed occurrences. We projected suitability shifts for the mid- (2041–2070) and end-of-century (2071–2100) periods under two CMIP6 Shared Socioeconomic Pathways (SSP3-7.0 and SSP5-8.5), using high-resolution CHELSA bioclimatic variables. The model, achieving an Area Under the Curve (AUC) of 0.73 and validated against 30 years of historical USDA production records, reveals a distinct spatial contraction of areas climatically suitable for growing maize. Projections indicate a significant decline in suitability across Western and Central Kansas driven by rising temperatures and precipitation variability, with the most highly suitable optimal habitats projected to decline by approximately 90% by mid-century. These findings quantify mounting climate impacts on maize-growing areas of the Great Plains and provide spatially explicit baselines for the development of regional adaptation strategies and groundwater conservation policies.

   @article{monavarian2026maize,
     title = {Machine Learning-Based Bioclimatic Suitability Modeling for Maize Cultivation Under Future Projections},
     author = {Monavarian, Alireza and Abadifard, Soheil and McGinty, Hande K. and Sharda, Vaishali},
     journal = {Land},
     volume = {15},
     number = {5},
     pages = {757},
     year = {2026},
     month = apr,
     publisher = {MDPI AG},
     issn = {2073-445X},
     doi = {10.3390/land15050757},
     url = {https://doi.org/10.3390/land15050757},
   }

2025

1. K-CAP

   GLIIDE: Global-Local Image Integration via Descriptive Extraction

   Aryan Singh Dalal, Soheil Abadifard, and Hande Küçük McGinty

   In Proceedings of the 13th Knowledge Capture Conference 2025 (K-CAP 2025), Dec 2025

   Talk at K-CAP 2025, Dayton, OH.

   Abs DOI Bib

   Automated Scene Graph Generation (SGG) is challenged by the "Global-Local Dilemma"—the tension between capturing holistic scene context and granular object details. This paper presents a preliminary investigation into a new generative SGG paradigm designed to address this challenge. We introduce a novel, two-track generative framework (GLIIDE) that creates a rich textual intermediary for a Large Language Model (LLM) to extract knowledge. Our hybrid pipeline fuses a YOLO-based track for extracting factual attributes with a DETR-guided track for generating contextual descriptions of object interactions. Our preliminary results on 1,000 images from the Visual Genome dataset demonstrate the potential of this fusion, which achieves a higher Relation Density than relative baselines. This work provides a preliminary proof-of-concept for a generative SGG paradigm that integrates local and global visual information.

   @inproceedings{dalal2025gliide,
     title = {GLIIDE: Global-Local Image Integration via Descriptive Extraction},
     author = {Dalal, Aryan Singh and Abadifard, Soheil and McGinty, Hande Küçük},
     booktitle = {Proceedings of the 13th Knowledge Capture Conference 2025 (K-CAP 2025)},
     note = {Talk at K-CAP 2025, Dayton, OH.},
     pages = {194--197},
     year = {2025},
     month = dec,
     publisher = {Association for Computing Machinery},
     address = {New York, NY, USA},
     isbn = {9798400718670},
     doi = {10.1145/3731443.3771367},
     url = {https://doi.org/10.1145/3731443.3771367},
   }

2. arXiv

   LSH-DynED: A Dynamic Ensemble Framework with LSH-Based Undersampling for Evolving Multi-Class Imbalanced Classification

   Soheil Abadifard and Fazli Can

   arXiv preprint arXiv:2506.20041, Jun 2025

   Abs arXiv Bib Code

   The classification of imbalanced data streams, which have unequal class distributions, is a key difficulty in machine learning, especially when dealing with multiple classes. While binary imbalanced data stream classification tasks have received considerable attention, only a few studies have focused on multi-class imbalanced data streams. Effectively managing the dynamic imbalance ratio is a key challenge in this domain. This study introduces a novel, robust, and resilient approach to address these challenges by integrating Locality Sensitive Hashing with Random Hyperplane Projections (LSH-RHP) into the Dynamic Ensemble Diversification (DynED) framework. To the best of our knowledge, we present the first application of LSH-RHP for undersampling in the context of imbalanced non-stationary data streams. The proposed method undersamples the majority classes by utilizing LSH-RHP, provides a balanced training set, and improves the ensemble’s prediction performance. We conduct comprehensive experiments on 23 real-world and ten semi-synthetic datasets and compare LSH-DynED with 15 state-of-the-art methods. The results reveal that LSH-DynED outperforms other approaches in terms of both Kappa and mG-Mean effectiveness measures, demonstrating its capability in dealing with multi-class imbalanced non-stationary data streams. Notably, LSH-DynED performs well in large-scale, high-dimensional datasets with considerable class imbalances and demonstrates adaptation and robustness in real-world circumstances. To motivate our design, we review existing methods for imbalanced data streams, outline key challenges, and offer guidance for future work. For the reproducibility of our results, we have made our implementation available on GitHub.

   @article{abadifard2025lshdyned,
     title = {LSH-DynED: A Dynamic Ensemble Framework with LSH-Based Undersampling for Evolving Multi-Class Imbalanced Classification},
     author = {Abadifard, Soheil and Can, Fazli},
     journal = {arXiv preprint arXiv:2506.20041},
     year = {2025},
     month = jun,
     archiveprefix = {arXiv},
     primaryclass = {cs.LG},
     url = {https://arxiv.org/abs/2506.20041},
   }

2024

1. Thesis

   Dynamic Ensemble Diversification and Hash-Based Undersampling for the Classification of Multi-Class Imbalanced Data Streams

   Soheil Abadifard

   Jul 2024

   M.Sc. thesis, Department of Computer Engineering. Advisor: Prof. Fazlı Can.

   Abs Bib HTML

   The classification of imbalanced data streams, which have unequal class distributions, is a key difficulty in machine learning, especially when dealing with multiple classes and concept drift. While binary imbalanced data stream classification tasks have received considerable attention, only a few studies have focused on multi-class imbalanced data streams. Additionally, dealing with the dynamic imbalance ratio is of great importance. This study introduces a novel, robust, and resilient approach to address these challenges by integrating Locality Sensitive Hashing with Random Hyperplane Projections (LSH-RHP) into the Dynamic Ensemble Diversification (DynED) framework. To the best of our knowledge, we present the first application of LSH-RHP for undersampling in the context of imbalanced non-stationary data streams. The proposed method, undersamples majority classes by utilizing LSH-RHP, provides a balanced training set, and improves the ensemble’s prediction accuracy. We conduct comprehensive experiments on 23 real-world and ten semi-synthetic datasets and compare LSH-DynED with 15 state-of-the-art methods. The results reveal that LSH-DynED outperforms other approaches in terms of both Kappa and mG-Mean effectiveness measures, demonstrating its capability in dealing with multi-class imbalanced non-stationary data streams. Notably, LSH-DynED performs well in large-scale, high-dimensional datasets with considerable class imbalances and demonstrates adaptation and robustness in real-world circumstances. For the reproducibility of our results, we have made our implementation available on GitHub.

   @mastersthesis{abadifard2024thesis,
     title = {Dynamic Ensemble Diversification and Hash-Based Undersampling for the Classification of Multi-Class Imbalanced Data Streams},
     author = {Abadifard, Soheil},
     school = {Bilkent University},
     type = {{M.S.} thesis},
     year = {2024},
     month = jul,
     note = {M.Sc. thesis, Department of Computer Engineering. Advisor: Prof. Fazl{\i} Can.},
     url = {https://hdl.handle.net/11693/115735},
   }

2023

1. CIKM

   DynED: Dynamic Ensemble Diversification in Data Stream Classification

   Soheil Abadifard, Sepehr Bakhshi, Sanaz Gheibuni, and Fazli Can

   In Proceedings of the 32nd ACM International Conference on Information and Knowledge Management (CIKM 2023), Oct 2023

   Poster at CIKM 2023, Birmingham, UK.

   Abs DOI Bib Code

   Ensemble methods are commonly used in classification due to their remarkable performance. Achieving high accuracy in a data stream environment is a challenging task considering disruptive changes in the data distribution, also known as concept drift. A greater diversity of ensemble components is known to enhance prediction accuracy in such settings. Despite the diversity of components within an ensemble, not all contribute as expected to its overall performance. This necessitates a method for selecting components that exhibit high performance and diversity. We present a novel ensemble construction and maintenance approach based on MMR (Maximal Marginal Relevance) that dynamically combines the diversity and prediction accuracy of components during the process of structuring an ensemble. The experimental results on both four real and 11 synthetic datasets demonstrate that the proposed approach (DynED) provides a higher average mean accuracy compared to the five state-of-the-art baselines.

   @inproceedings{abadifard2023dyned,
     title = {DynED: Dynamic Ensemble Diversification in Data Stream Classification},
     author = {Abadifard, Soheil and Bakhshi, Sepehr and Gheibuni, Sanaz and Can, Fazli},
     booktitle = {Proceedings of the 32nd ACM International Conference on Information and Knowledge Management (CIKM 2023)},
     note = {Poster at CIKM 2023, Birmingham, UK.},
     pages = {3707--3711},
     year = {2023},
     month = oct,
     publisher = {Association for Computing Machinery},
     address = {New York, NY, USA},
     isbn = {9798400701245},
     doi = {10.1145/3583780.3615266},
     url = {https://doi.org/10.1145/3583780.3615266},
   }