arXiv subjects

Ashley Suh

Publications and source records attributed to Ashley Suh.

18 recordsLinked to original sources

arXiv2025

The Role of Visualization in LLM-Assisted Knowledge Graph Systems: Effects on User Trust, Exploration, and Workflows

Knowledge graphs (KGs) are powerful data structures, but exploring them effectively remains difficult for even expert users. Large language models (LLMs) are increasingly used to address this gap, yet little is known empirically about how their usage with KGs shapes user trust, exploration strategies, or downstream decision-making - raising key design challenges for LLM-based KG visual analysis systems. To study these effects, we developed LinkQ, a KG exploration system that converts natural language questions into structured queries with an LLM. We collaborated with KG experts to design five visual mechanisms that help users assess the accuracy of both KG queries and LLM responses: an LLM-KG state diagram that illustrates which stage of the exploration pipeline LinkQ is in, a query editor displaying the generated query paired with an LLM explanation, an entity-relation ID table showing extracted KG entities and relations with semantic descriptions, a query structure graph that depicts the path traversed in the KG, and an interactive graph visualization of query results. From a qualitative evaluation with 14 practitioners, we found that users - even KG experts - tended to overtrust LinkQ's outputs due to its "helpful" visualizations, even when the LLM was incorrect. Users exhibited distinct workflows depending on their prior familiarity with KGs and LLMs, challenging the assumption that these systems are one-size-fits-all - despite often being designed as if they are. Our findings highlight the risks of false trust in LLM-assisted data analysis tools and the need for further investigation into the role of visualization as a mitigation technique.

cs.LG↗

arXiv2025

Mitigating LLM Hallucinations with Knowledge Graphs: A Case Study

High-stakes domains like cyber operations need responsible and trustworthy AI methods. While large language models (LLMs) are becoming increasingly popular in these domains, they still suffer from hallucinations. This research paper provides learning outcomes from a case study with LinkQ, an open-source natural language interface that was developed to combat hallucinations by forcing an LLM to query a knowledge graph (KG) for ground-truth data during question-answering (QA). We conduct a quantitative evaluation of LinkQ using a well-known KGQA dataset, showing that the system outperforms GPT-4 but still struggles with certain question categories - suggesting that alternative query construction strategies will need to be investigated in future LLM querying systems. We discuss a qualitative study of LinkQ with two domain experts using a real-world cybersecurity KG, outlining these experts' feedback, suggestions, perceived limitations, and future opportunities for systems like LinkQ.

cs.HC↗

arXiv2025

Don't Just Translate, Agitate: Using Large Language Models as Devil's Advocates for AI Explanations

This position paper highlights a growing trend in Explainable AI (XAI) research where Large Language Models (LLMs) are used to translate outputs from explainability techniques, like feature-attribution weights, into a natural language explanation. While this approach may improve accessibility or readability for users, recent findings suggest that translating into human-like explanations does not necessarily enhance user understanding and may instead lead to overreliance on AI systems. When LLMs summarize XAI outputs without surfacing model limitations, uncertainties, or inconsistencies, they risk reinforcing the illusion of interpretability rather than fostering meaningful transparency. We argue that - instead of merely translating XAI outputs - LLMs should serve as constructive agitators, or devil's advocates, whose role is to actively interrogate AI explanations by presenting alternative interpretations, potential biases, training data limitations, and cases where the model's reasoning may break down. In this role, LLMs can facilitate users in engaging critically with AI systems and generated explanations, with the potential to reduce overreliance caused by misinterpreted or specious explanations.

cs.HC↗

arXiv2025

Fewer Than 1% of Explainable AI Papers Validate Explainability with Humans

This late-breaking work presents a large-scale analysis of explainable AI (XAI) literature to evaluate claims of human explainability. We collaborated with a professional librarian to identify 18,254 papers containing keywords related to explainability and interpretability. Of these, we find that only 253 papers included terms suggesting human involvement in evaluating an XAI technique, and just 128 of those conducted some form of a human study. In other words, fewer than 1% of XAI papers (0.7%) provide empirical evidence of human explainability when compared to the broader body of XAI literature. Our findings underscore a critical gap between claims of human explainability and evidence-based validation, raising concerns about the rigor of XAI research. We call for increased emphasis on human evaluations in XAI studies and provide our literature search methodology to enable both reproducibility and further investigation into this widespread issue.

cs.HC↗

arXiv2024

STL: Still Tricky Logic (for System Validation, Even When Showing Your Work)

As learned control policies become increasingly common in autonomous systems, there is increasing need to ensure that they are interpretable and can be checked by human stakeholders. Formal specifications have been proposed as ways to produce human-interpretable policies for autonomous systems that can still be learned from examples. Previous work showed that despite claims of interpretability, humans are unable to use formal specifications presented in a variety of ways to validate even simple robot behaviors. This work uses active learning, a standard pedagogical method, to attempt to improve humans' ability to validate policies in signal temporal logic (STL). Results show that overall validation accuracy is not high, at $65\% \pm 15\%$ (mean $\pm$ standard deviation), and that the three conditions of no active learning, active learning, and active learning with feedback do not significantly differ from each other. Our results suggest that the utility of formal specifications for human interpretability is still unsupported but point to other avenues of development which may enable improvements in system validation.

cs.HC↗

arXiv2024

LinkQ: An LLM-Assisted Visual Interface for Knowledge Graph Question-Answering

We present LinkQ, a system that leverages a large language model (LLM) to facilitate knowledge graph (KG) query construction through natural language question-answering. Traditional approaches often require detailed knowledge of a graph querying language, limiting the ability for users -- even experts -- to acquire valuable insights from KGs. LinkQ simplifies this process by implementing a multistep protocol in which the LLM interprets a user's question, then systematically converts it into a well-formed query. LinkQ helps users iteratively refine any open-ended questions into precise ones, supporting both targeted and exploratory analysis. Further, LinkQ guards against the LLM hallucinating outputs by ensuring users' questions are only ever answered from ground truth KG data. We demonstrate the efficacy of LinkQ through a qualitative study with five KG practitioners. Our results indicate that practitioners find LinkQ effective for KG question-answering, and desire future LLM-assisted exploratory data analysis systems.

cs.CL↗

arXiv2024

A Preliminary Roadmap for LLMs as Assistants in Exploring, Analyzing, and Visualizing Knowledge Graphs

We present a mixed-methods study to explore how large language models (LLMs) can assist users in the visual exploration and analysis of knowledge graphs (KGs). We surveyed and interviewed 20 professionals from industry, government laboratories, and academia who regularly work with KGs and LLMs, either collaboratively or concurrently. Our findings show that participants overwhelmingly want an LLM to facilitate data retrieval from KGs through joint query construction, to identify interesting relationships in the KG through multi-turn conversation, and to create on-demand visualizations from the KG that enhance their trust in the LLM's outputs. To interact with an LLM, participants strongly prefer a chat-based 'widget,' built on top of their regular analysis workflows, with the ability to guide the LLM using their interactions with a visualization. When viewing an LLM's outputs, participants similarly prefer a combination of annotated visuals (e.g., subgraphs or tables extracted from the KG) alongside summarizing text. However, participants also expressed concerns about an LLM's ability to maintain semantic intent when translating natural language questions into KG queries, the risk of an LLM 'hallucinating' false data from the KG, and the difficulties of engineering a 'perfect prompt.' From the analysis of our interviews, we contribute a preliminary roadmap for the design of LLM-driven knowledge graph exploration systems and outline future opportunities in this emergent design space.

cs.HC↗

arXiv2024

Preliminary Guidelines For Combining Data Integration and Visual Data Analysis

Data integration is often performed to consolidate information from multiple disparate data sources during visual data analysis. However, integration operations are usually separate from visual analytics operations such as encode and filter in both interface design and empirical research. We conducted a preliminary user study to investigate whether and how data integration should be incorporated directly into the visual analytics process. We used two interface alternatives featuring contrasting approaches to the data preparation and analysis workflow: manual file-based ex-situ integration as a separate step from visual analytics operations; and automatic UI-based in-situ integration merged with visual analytics operations. Participants were asked to complete specific and free-form tasks with each interface, browsing for patterns, generating insights, and summarizing relationships between attributes distributed across multiple files. Analyzing participants' interactions and feedback, we found both task completion time and total interactions to be similar across interfaces and tasks, as well as unique integration strategies between interfaces and emergent behaviors related to satisficing and cognitive bias. Participants' time spent and interactions revealed that in-situ integration enabled users to spend more time on analysis tasks compared with ex-situ integration. Participants' integration strategies and analytical behaviors revealed differences in interface usage for generating and tracking hypotheses and insights. With these results, we synthesized preliminary guidelines for designing future visual analytics interfaces that can support integrating attributes throughout an active analysis process.

cs.HC↗

arXiv2024

Visual Validation versus Visual Estimation: A Study on the Average Value in Scatterplots

We investigate the ability of individuals to visually validate statistical models in terms of their fit to the data. While visual model estimation has been studied extensively, visual model validation remains under-investigated. It is unknown how well people are able to visually validate models, and how their performance compares to visual and computational estimation. As a starting point, we conducted a study across two populations (crowdsourced and volunteers). Participants had to both visually estimate (i.e, draw) and visually validate (i.e., accept or reject) the frequently studied model of averages. Across both populations, the level of accuracy of the models that were considered valid was lower than the accuracy of the estimated models. We find that participants' validation and estimation were unbiased. Moreover, their natural critical point between accepting and rejecting a given mean value is close to the boundary of its 95% confidence interval, indicating that the visually perceived confidence interval corresponds to a common statistical standard. Our work contributes to the understanding of visual model validation and opens new research opportunities.

cs.CV↗

arXiv2023

A Grammar of Hypotheses for Visualization, Data, and Analysis

We present a grammar for expressing hypotheses in visual data analysis to formalize the previously abstract notion of "analysis tasks." Through the lens of our grammar, we lay the groundwork for how a user's data analysis questions can be operationalized and automated as a set of hypotheses (a hypothesis space). We demonstrate that our grammar-based approach for analysis tasks can provide a systematic method towards unifying three disparate spaces in visualization research: the hypotheses a dataset can express (a data hypothesis space), the hypotheses a user would like to refine or verify through analysis (an analysis hypothesis space), and the hypotheses a visualization design is capable of supporting (a visualization hypothesis space). We illustrate how the formalization of these three spaces can inform future research in visualization evaluation, knowledge elicitation, analytic provenance, and visualization recommendation by using a shared language for hypotheses. Finally, we compare our proposed grammar-based approach with existing visual analysis models and discuss the potential of a new hypothesis-driven theory of visual analytics.

cs.HC↗

arXiv2023

Knowledge Graphs in Practice: Characterizing their Users, Challenges, and Visualization Opportunities

This study presents insights from interviews with nineteen Knowledge Graph (KG) practitioners who work in both enterprise and academic settings on a wide variety of use cases. Through this study, we identify critical challenges experienced by KG practitioners when creating, exploring, and analyzing KGs that could be alleviated through visualization design. Our findings reveal three major personas among KG practitioners - KG Builders, Analysts, and Consumers - each of whom have their own distinct expertise and needs. We discover that KG Builders would benefit from schema enforcers, while KG Analysts need customizable query builders that provide interim query results. For KG Consumers, we identify a lack of efficacy for node-link diagrams, and the need for tailored domain-specific visualizations to promote KG adoption and comprehension. Lastly, we find that implementing KGs effectively in practice requires both technical and social solutions that are not addressed with current tools, technologies, and collaborative workflows. From the analysis of our interviews, we distill several visualization research directions to improve KG usability, including knowledge cards that balance digestibility and discoverability, timeline views to track temporal changes, interfaces that support organic discovery, and semantic explanations for AI and machine learning predictions.

cs.HC↗

arXiv2022

Analysis Without Data: Teaching Students to Tackle the VAST Challenge

The VAST Challenges have been shown to be an effective tool in visual analytics education, encouraging student learning while enforcing good visualization design and development practices. However, research has observed that students often struggle at identifying a good "starting point" when tackling the VAST Challenge. Consequently, students who could not identify a good starting point failed at finding the correct solution to the challenge. In this paper, we propose a preliminary guideline for helping students approach the VAST Challenge and identify initial analysis directions. We recruited two students to analyze the VAST 2017 Challenge using a hypothesis-driven approach, where they were required to pre-register their hypotheses prior to inspecting and analyzing the full dataset. From their experience, we developed a prescriptive guideline for other students to tackle VAST Challenges. In a preliminary study, we found that the students were able to use the guideline to generate well-formed hypotheses that could lead them towards solving the challenge. Additionally, the students reported that with the guideline, they felt like they had concrete steps that they could follow, thereby alleviating the burden of identifying a good starting point in their analysis process.

cs.HC↗

arXiv2022

Inferential Tasks as an Evaluation Technique for Visualization

Designing suitable tasks for visualization evaluation remains challenging. Traditional evaluation techniques commonly rely on 'low-level' or 'open-ended' tasks to assess the efficacy of a proposed visualization, however, nontrivial trade-offs exist between the two. Low-level tasks allow for robust quantitative evaluations, but are not indicative of the complex usage of a visualization. Open-ended tasks, while excellent for insight-based evaluations, are typically unstructured and require time-consuming interviews. Bridging this gap, we propose inferential tasks: a complementary task category based on inferential learning in psychology. Inferential tasks produce quantitative evaluation data in which users are prompted to form and validate their own findings with a visualization. We demonstrate the use of inferential tasks through a validation experiment on two well-known visualization tools.

cs.HC↗

arXiv2022

Are Metrics Enough? Guidelines for Communicating and Visualizing Predictive Models to Subject Matter Experts

Presenting a predictive model's performance is a communication bottleneck that threatens collaborations between data scientists and subject matter experts. Accuracy and error metrics alone fail to tell the whole story of a model - its risks, strengths, and limitations - making it difficult for subject matter experts to feel confident in their decision to use a model. As a result, models may fail in unexpected ways or go entirely unused, as subject matter experts disregard poorly presented models in favor of familiar, yet arguably substandard methods. In this paper, we describe an iterative study conducted with both subject matter experts and data scientists to understand the gaps in communication between these two groups. We find that, while the two groups share common goals of understanding the data and predictions of the model, friction can stem from unfamiliar terms, metrics, and visualizations - limiting the transfer of knowledge to SMEs and discouraging clarifying questions being asked during presentations. Based on our findings, we derive a set of communication guidelines that use visualization as a common medium for communicating the strengths and weaknesses of a model. We provide a demonstration of our guidelines in a regression modeling scenario and elicit feedback on their use from subject matter experts. From our demonstration, subject matter experts were more comfortable discussing a model's performance, more aware of the trade-offs for the presented model, and better equipped to assess the model's risks - ultimately informing and contextualizing the model's use beyond text and numbers.

cs.HC↗

arXiv2021

UnProjection: Leveraging Inverse-Projections for Visual Analytics of High-Dimensional Data

Projection techniques are often used to visualize high-dimensional data, allowing users to better understand the overall structure of multi-dimensional spaces on a 2D screen. Although many such methods exist, comparably little work has been done on generalizable methods of inverse-projection -- the process of mapping the projected points, or more generally, the projection space back to the original high-dimensional space. In this paper we present NNInv, a deep learning technique with the ability to approximate the inverse of any projection or mapping. NNInv learns to reconstruct high-dimensional data from any arbitrary point on a 2D projection space, giving users the ability to interact with the learned high-dimensional representation in a visual analytics system. We provide an analysis of the parameter space of NNInv, and offer guidance in selecting these parameters. We extend validation of the effectiveness of NNInv through a series of quantitative and qualitative analyses. We then demonstrate the method's utility by applying it to three visualization tasks: interactive instance interpolation, classifier agreement, and gradient visualization.

cs.HC↗

arXiv2020

TopoLines: Topological Smoothing for Line Charts

Line charts are commonly used to visualize a series of data values. When the data are noisy, smoothing is applied to make the signal more apparent. Conventional methods used to smooth line charts, e.g., using subsampling or filters, such as median, Gaussian, or low-pass, each optimize for different properties of the data. The properties generally do not include retaining peaks (i.e., local minima and maxima) in the data, which is an important feature for certain visual analytics tasks. We present TopoLines, a method for smoothing line charts using techniques from Topological Data Analysis. The design goal of TopoLines is to maintain prominent peaks in the data while minimizing any residual error. We evaluate TopoLines for 2 visual analytics tasks by comparing to 5 popular line smoothing methods with data from 4 application domains.

cs.HC↗

arXiv2019

Persistent Homology Guided Force-Directed Graph Layouts

Graphs are commonly used to encode relationships among entities, yet their abstractness makes them difficult to analyze. Node-link diagrams are popular for drawing graphs, and force-directed layouts provide a flexible method for node arrangements that use local relationships in an attempt to reveal the global shape of the graph. However, clutter and overlap of unrelated structures can lead to confusing graph visualizations. This paper leverages the persistent homology features of an undirected graph as derived information for interactive manipulation of force-directed layouts. We first discuss how to efficiently extract 0-dimensional persistent homology features from both weighted and unweighted undirected graphs. We then introduce the interactive persistence barcode used to manipulate the force-directed graph layout. In particular, the user adds and removes contracting and repulsing forces generated by the persistent homology features, eventually selecting the set of persistent homology features that most improve the layout. Finally, we demonstrate the utility of our approach across a variety of synthetic and real datasets.

cs.GR↗