arXiv subjects

Tsvetomila Mihaylova

Publications and source records attributed to Tsvetomila Mihaylova.

15 recordsLinked to original sources

arXiv2026

When Looks Do Not Lie: Discourse Structure Guided In-Context Learning for Faithful Diagram Generation

GenAI is widespread in educational applications; however, it is known to generate content with intrinsic and extrinsic hallucination. We introduce a novel method for ICL diagram generation based on Rhetorical Structure Theory, which improves diagram faithfulness to its source text context. We find that ICL performance depends on task distribution and models' reasoning ability, with higher reasoning allowing better quality and performance for an out-of-distribution task. We perform an expert evaluation of 150 generated diagrams and analyze our findings using Bayesian GLMMs. Additionally, we use our evaluation rubric and samples from the data set for automated diagram evaluation, achieving statistically significant agreement with human evaluation.

cs.CL↗

arXiv2026

Relational Scene Graphs for Object Grounding of Natural Language Commands

Robots are finding wider adoption in human environments, increasing the need for natural human-robot interaction. However, understanding a natural language command requires the robot to infer the intended task and how to decompose it into executable actions, and to ground those actions in the robot's knowledge of the environment, including relevant objects, agents, and locations. This challenge can be addressed by combining the capabilities of large language models (LLMs) to understand natural language with 3D scene graphs (3DSGs) for grounding inferred actions in a semantic representation of the environment. However, many 3DSGs lack explicit spatial relations between objects, even though humans often rely on these relations to describe an environment. This paper investigates whether incorporating open- or closed-vocabulary spatial relations into 3DSGs can improve the ability of LLMs to interpret natural language commands. To address this, we implement two pipelines using off-the-shelf models: an LLM-based pipeline for target object grounding from open-vocabulary language commands and a vision language model-based pipeline to add open-vocabulary spatial edges to 3DSGs from images captured while mapping. Finally, we evaluate two LLMs across 14 scenes using 905 natural language statements (786 procedurally-generated, 119 human-authored) to assess performance on the downstream task of target object grounding. Our study demonstrates that explicit spatial relations improve the ability of LLMs to ground objects, and while open-vocabulary relation generation with vision language models proves feasible from robot-captured images, our analysis did not yield evidence favoring either open- or closed-vocabulary relations.

cs.RO↗

arXiv2026

Analyzing the Difficulty of Programming Assignments with Interpretable Knowledge Component Metrics

This research paper examines how Knowledge Components (KCs) - fine-grained concepts or skills required to solve programming tasks - can be used as interpretable signals for understanding assignment difficulty and student struggle in introductory programming courses. While prior work has focused on predictive models based on programming behavior, such models are often difficult to interpret and therefore hard to use for instructional decisions. We analyze KC-based metrics, including the number of KCs per assignment and changes in KC coverage between consecutive assignments. We examine correlations between the number of KCs and student performance on the assignment, and analyze changes in KCs across assignments to identify cases where performance declines without new concepts being introduced. Selected assignments are then qualitatively inspected to understand potential design issues. Our results on data from three introductory programming course datasets show that assignments involving more KCs are generally associated with lower performance, and that sudden shifts in required KCs can coincide with disruptions in learning progression. We also identify assignments where performance declines even though no new KCs are introduced, suggesting potential issues in task design or instruction, which could be examined with qualitative analysis. We propose an interpretable framework for analyzing programming assignments using KC-based metrics, with practical implications for instructors and course designers who want to better understand where and why students struggle, and how course materials might be improved. Our method can use KCs defined by an expert or extracted by an LLM, offering instructors an additional way to assess assignment quality beyond average correctness. It can be applied to any course with ordered assignments and measurable performance.

cs.CY↗

arXiv2026

The Effects of Structured LLM-Generated Feedback on Programming Assignment Performance

When programming students encounter errors in their code, compiler messages or static analysis output often provide limited guidance, particularly for novice programmers. Personalized feedback from instructors can be effective but does not scale well. Recent advances in large language models (LLMs) enable automated feedback generation at scale. This study examines whether LLM-generated feedback with different levels of guidance is associated with differences in students' problem-solving behavior. We analyze effects on time to solution and number of attempts, and examine whether these effects differ by programming experience. We design three feedback types and compare them to a baseline in which students receive only compiler error messages. Results from an online programming course show that LLM-generated feedback is associated with faster time to solution compared to the no-feedback baseline, with less guided feedback showing slightly stronger effects. Overall, the findings suggest that feedback structure plays an important role in how students progress toward correct solutions and motivate further work on adaptive feedback designs and longer-term learning outcomes.

cs.HC↗

arXiv2026

Discrete Latent Structure in Neural Networks

Many types of data from fields including natural language processing, computer vision, and bioinformatics, are well represented by discrete, compositional structures such as trees, sequences, or matchings. Latent structure models are a powerful tool for learning to extract such representations, offering a way to incorporate structural bias, discover insight about the data, and interpret decisions. However, effective training is challenging, as neural networks are typically designed for continuous computation. This text explores three broad strategies for learning with discrete latent structure: continuous relaxation, surrogate gradients, and probabilistic estimation. Our presentation relies on consistent notations for a wide range of models. As such, we reveal many new connections between latent structure learning strategies, showing how most consist of the same small set of fundamental building blocks, but use them differently, leading to substantially different applicability and properties.

cs.LG↗

arXiv2025

Do Visual-Language Grid Maps Capture Latent Semantics?

Visual-language models (VLMs) have recently been introduced in robotic mapping using the latent representations, i.e., embeddings, of the VLMs to represent semantics in the map. They allow moving from a limited set of human-created labels toward open-vocabulary scene understanding, which is very useful for robots when operating in complex real-world environments and interacting with humans. While there is anecdotal evidence that maps built this way support downstream tasks, such as navigation, rigorous analysis of the quality of the maps using these embeddings is missing. In this paper, we propose a way to analyze the quality of maps created using VLMs. We investigate two critical properties of map quality: queryability and distinctness. The evaluation of queryability addresses the ability to retrieve information from the embeddings. We investigate intra-map distinctness to study the ability of the embeddings to represent abstract semantic classes and inter-map distinctness to evaluate the generalization properties of the representation. We propose metrics to evaluate these properties and evaluate two state-of-the-art mapping methods, VLMaps and OpenScene, using two encoders, LSeg and OpenSeg, using real-world data from the Matterport3D data set. Our findings show that while 3D features improve queryability, they are not scale invariant, whereas image-based embeddings generalize to multiple map resolutions. This allows the image-based methods to maintain smaller map sizes, which can be crucial for using these methods in real-world deployments. Furthermore, we show that the choice of the encoder has an effect on the results. The results imply that properly thresholding open-vocabulary queries is an open problem.

cs.RO↗

arXiv2025

Injecting Conflict Situations in Autonomous Driving Simulation using CARLA

Simulation of conflict situations for autonomous driving research is crucial for understanding and managing interactions between Automated Vehicles (AVs) and human drivers. This paper presents a set of exemplary conflict scenarios in CARLA that arise in shared autonomy settings, where both AVs and human drivers must navigate complex traffic environments. We explore various conflict situations, focusing on the impact of driver behavior and decision-making processes on overall traffic safety and efficiency. We build a simple extendable toolkit for situation awareness research, in which the implemented conflicts can be demonstrated.

cs.HC↗

arXiv2022

Modeling Structure with Undirected Neural Networks

Neural networks are powerful function estimators, leading to their status as a paradigm of choice for modeling structured data. However, unlike other structured representations that emphasize the modularity of the problem -- e.g., factor graphs -- neural networks are usually monolithic mappings from inputs to outputs, with a fixed computation order. This limitation prevents them from capturing different directions of computation and interaction between the modeled variables. In this paper, we combine the representational strengths of factor graphs and of neural networks, proposing undirected neural networks (UNNs): a flexible framework for specifying computations that can be performed in any order. For particular choices, our proposed models subsume and extend many existing architectures: feed-forward, recurrent, self-attention networks, auto-encoders, and networks with implicit layers. We demonstrate the effectiveness of undirected neural architectures, both unstructured and structured, on a range of tasks: tree-constrained dependency parsing, convolutional image classification, and sequence completion with attention. By varying the computation order, we show how a single UNN can be used both as a classifier and a prototype generator, and how it can fill in missing parts of an input sequence, making them a promising field for further research.

cs.LG↗

arXiv2021

SUper Team at SemEval-2016 Task 3: Building a feature-rich system for community question answering

We present the system we built for participating in SemEval-2016 Task 3 on Community Question Answering. We achieved the best results on subtask C, and strong results on subtasks A and B, by combining a rich set of various types of features: semantic, lexical, metadata, and user-related. The most important group turned out to be the metadata for the question and for the comment, semantic vectors trained on QatarLiving data and similarities between the question and the comment for subtasks A and C, and between the original and the related question for Subtask B.

cs.CL↗

arXiv2020

Understanding the Mechanics of SPIGOT: Surrogate Gradients for Latent Structure Learning

Latent structure models are a powerful tool for modeling language data: they can mitigate the error propagation and annotation bottleneck in pipeline systems, while simultaneously uncovering linguistic insights about the data. One challenge with end-to-end training of these models is the argmax operation, which has null gradient. In this paper, we focus on surrogate gradients, a popular strategy to deal with this problem. We explore latent structure learning through the angle of pulling back the downstream learning objective. In this paradigm, we discover a principled motivation for both the straight-through estimator (STE) as well as the recently-proposed SPIGOT - a variant of STE for structured models. Our perspective leads to new algorithms in the same family. We empirically compare the known and the novel pulled-back estimators against the popular alternatives, yielding new insight for practitioners and revealing intriguing failure cases.

cs.CL↗

arXiv2019

Automatic Fact-Checking Using Context and Discourse Information

We study the problem of automatic fact-checking, paying special attention to the impact of contextual and discourse information. We address two related tasks: (i) detecting check-worthy claims, and (ii) fact-checking claims. We develop supervised systems based on neural networks, kernel-based support vector machines, and combinations thereof, which make use of rich input representations in terms of discourse cues and contextual features. For the check-worthiness estimation task, we focus on political debates, and we model the target claim in the context of the full intervention of a participant and the previous and the following turns in the debate, taking into account contextual meta information. For the fact-checking task, we focus on answer verification in a community forum, and we model the veracity of the answer with respect to the entire question--answer thread in which it occurs as well as with respect to other related posts from the entire forum. We develop annotated datasets for both tasks and we run extensive experimental evaluation, confirming that both types of information ---but especially contextual features--- play an important role.

cs.CL↗

arXiv2019

Scheduled Sampling for Transformers

Scheduled sampling is a technique for avoiding one of the known problems in sequence-to-sequence generation: exposure bias. It consists of feeding the model a mix of the teacher forced embeddings and the model predictions from the previous step in training time. The technique has been used for improving the model performance with recurrent neural networks (RNN). In the Transformer model, unlike the RNN, the generation of a new word attends to the full sentence generated so far, not only to the last word, and it is not straightforward to apply the scheduled sampling technique. We propose some structural changes to allow scheduled sampling to be applied to Transformer architecture, via a two-pass decoding strategy. Experiments on two language pairs achieve performance close to a teacher-forcing baseline and show that this technique is promising for further exploration.

cs.CL↗

arXiv2019

SemEval-2019 Task 8: Fact Checking in Community Question Answering Forums

We present SemEval-2019 Task 8 on Fact Checking in Community Question Answering Forums, which features two subtasks. Subtask A is about deciding whether a question asks for factual information vs. an opinion/advice vs. just socializing. Subtask B asks to predict whether an answer to a factual question is true, false or not a proper answer. We received 17 official submissions for subtask A and 11 official submissions for Subtask B. For subtask A, all systems improved over the majority class baseline. For Subtask B, all systems were below a majority class baseline, but several systems were very close to it. The leaderboard and the data from the competition can be found at http://competitions.codalab.org/competitions/20022

cs.CL↗

arXiv2018

Fact Checking in Community Forums

Community Question Answering (cQA) forums are very popular nowadays, as they represent effective means for communities around particular topics to share information. Unfortunately, this information is not always factual. Thus, here we explore a new dimension in the context of cQA, which has been ignored so far: checking the veracity of answers to particular questions in cQA forums. As this is a new problem, we create a specialized dataset for it. We further propose a novel multi-faceted model, which captures information from the answer content (what is said and how), from the author profile (who says it), from the rest of the community forum (where it is said), and from external authoritative sources of information (external support). Evaluation results show a MAP value of 86.54, which is 21 points absolute above the baseline.

cs.CL↗

arXiv2017

The Case for Being Average: A Mediocrity Approach to Style Masking and Author Obfuscation

Users posting online expect to remain anonymous unless they have logged in, which is often needed for them to be able to discuss freely on various topics. Preserving the anonymity of a text's writer can be also important in some other contexts, e.g., in the case of witness protection or anonymity programs. However, each person has his/her own style of writing, which can be analyzed using stylometry, and as a result, the true identity of the author of a piece of text can be revealed even if s/he has tried to hide it. Thus, it could be helpful to design automatic tools that can help a person obfuscate his/her identity when writing text. In particular, here we propose an approach that changes the text, so that it is pushed towards average values for some general stylometric characteristics, thus making the use of these characteristics less discriminative. The approach consists of three main steps: first, we calculate the values for some popular stylometric metrics that can indicate authorship; then we apply various transformations to the text, so that these metrics are adjusted towards the average level, while preserving the semantics and the soundness of the text; and finally, we add random noise. This approach turned out to be very efficient, and yielded the best performance on the Author Obfuscation task at the PAN-2016 competition.

cs.CL↗