-
Understanding Common Ground Misalignment in Goal-Oriented Dialog: A Case-Study with Ubuntu Chat Logs
Authors:
Rupak Sarkar,
Neha Srikanth,
Taylor Hudson,
Rachel Rudinger,
Claire Bonial,
Philip Resnik
Abstract:
While it is commonly accepted that maintaining common ground plays a role in conversational success, little prior research exists connecting conversational grounding to success in task-oriented conversations. We study failures of grounding in the Ubuntu IRC dataset, where participants use text-only communication to resolve technical issues. We find that disruptions in conversational flow often ste…
▽ More
While it is commonly accepted that maintaining common ground plays a role in conversational success, little prior research exists connecting conversational grounding to success in task-oriented conversations. We study failures of grounding in the Ubuntu IRC dataset, where participants use text-only communication to resolve technical issues. We find that disruptions in conversational flow often stem from a misalignment in common ground, driven by a divergence in beliefs and assumptions held by participants. These disruptions, which we call conversational friction, significantly correlate with task success. We find that although LLMs can identify overt cases of conversational friction, they struggle with subtler and more context-dependent instances requiring pragmatic or domain-specific reasoning.
△ Less
Submitted 16 March, 2025;
originally announced March 2025.
-
NLI under the Microscope: What Atomic Hypothesis Decomposition Reveals
Authors:
Neha Srikanth,
Rachel Rudinger
Abstract:
Decomposition of text into atomic propositions is a flexible framework allowing for the closer inspection of input and output text. We use atomic decomposition of hypotheses in two natural language reasoning tasks, traditional NLI and defeasible NLI, to form atomic sub-problems, or granular inferences that models must weigh when solving the overall problem. These atomic sub-problems serve as a too…
▽ More
Decomposition of text into atomic propositions is a flexible framework allowing for the closer inspection of input and output text. We use atomic decomposition of hypotheses in two natural language reasoning tasks, traditional NLI and defeasible NLI, to form atomic sub-problems, or granular inferences that models must weigh when solving the overall problem. These atomic sub-problems serve as a tool to further understand the structure of both NLI and defeasible reasoning, probe a model's consistency and understanding of different inferences, and measure the diversity of examples in benchmark datasets. Our results indicate that LLMs still struggle with logical consistency on atomic NLI and defeasible NLI sub-problems. Lastly, we identify critical atomic sub-problems of defeasible NLI examples, or those that most contribute to the overall label, and propose a method to measure the inferential consistency of a model, a metric designed to capture the degree to which a model makes consistently correct or incorrect predictions about the same fact under different contexts.
△ Less
Submitted 7 March, 2025; v1 submitted 11 February, 2025;
originally announced February 2025.
-
How often are errors in natural language reasoning due to paraphrastic variability?
Authors:
Neha Srikanth,
Marine Carpuat,
Rachel Rudinger
Abstract:
Large language models have been shown to behave inconsistently in response to meaning-preserving paraphrastic inputs. At the same time, researchers evaluate the knowledge and reasoning abilities of these models with test evaluations that do not disaggregate the effect of paraphrastic variability on performance. We propose a metric for evaluating the paraphrastic consistency of natural language rea…
▽ More
Large language models have been shown to behave inconsistently in response to meaning-preserving paraphrastic inputs. At the same time, researchers evaluate the knowledge and reasoning abilities of these models with test evaluations that do not disaggregate the effect of paraphrastic variability on performance. We propose a metric for evaluating the paraphrastic consistency of natural language reasoning models based on the probability of a model achieving the same correctness on two paraphrases of the same problem. We mathematically connect this metric to the proportion of a model's variance in correctness attributable to paraphrasing. To estimate paraphrastic consistency, we collect ParaNLU, a dataset of 7,782 human-written and validated paraphrased reasoning problems constructed on top of existing benchmark datasets for defeasible and abductive natural language inference. Using ParaNLU, we measure the paraphrastic consistency of several model classes and show that consistency dramatically increases with pretraining but not finetuning. All models tested exhibited room for improvement in paraphrastic consistency.
△ Less
Submitted 17 April, 2024;
originally announced April 2024.
-
Pregnant Questions: The Importance of Pragmatic Awareness in Maternal Health Question Answering
Authors:
Neha Srikanth,
Rupak Sarkar,
Heran Mane,
Elizabeth M. Aparicio,
Quynh C. Nguyen,
Rachel Rudinger,
Jordan Boyd-Graber
Abstract:
Questions posed by information-seeking users often contain implicit false or potentially harmful assumptions. In a high-risk domain such as maternal and infant health, a question-answering system must recognize these pragmatic constraints and go beyond simply answering user questions, examining them in context to respond helpfully. To achieve this, we study assumptions and implications, or pragmat…
▽ More
Questions posed by information-seeking users often contain implicit false or potentially harmful assumptions. In a high-risk domain such as maternal and infant health, a question-answering system must recognize these pragmatic constraints and go beyond simply answering user questions, examining them in context to respond helpfully. To achieve this, we study assumptions and implications, or pragmatic inferences, made when mothers ask questions about pregnancy and infant care by collecting a dataset of 2,727 inferences from 500 questions across three diverse sources. We study how health experts naturally address these inferences when writing answers, and illustrate that informing existing QA pipelines with pragmatic inferences produces responses that are more complete, mitigating the propagation of harmful beliefs.
△ Less
Submitted 2 April, 2024; v1 submitted 15 November, 2023;
originally announced November 2023.
-
HyHTM: Hyperbolic Geometry based Hierarchical Topic Models
Authors:
Simra Shahid,
Tanay Anand,
Nikitha Srikanth,
Sumit Bhatia,
Balaji Krishnamurthy,
Nikaash Puri
Abstract:
Hierarchical Topic Models (HTMs) are useful for discovering topic hierarchies in a collection of documents. However, traditional HTMs often produce hierarchies where lowerlevel topics are unrelated and not specific enough to their higher-level topics. Additionally, these methods can be computationally expensive. We present HyHTM - a Hyperbolic geometry based Hierarchical Topic Models - that addres…
▽ More
Hierarchical Topic Models (HTMs) are useful for discovering topic hierarchies in a collection of documents. However, traditional HTMs often produce hierarchies where lowerlevel topics are unrelated and not specific enough to their higher-level topics. Additionally, these methods can be computationally expensive. We present HyHTM - a Hyperbolic geometry based Hierarchical Topic Models - that addresses these limitations by incorporating hierarchical information from hyperbolic geometry to explicitly model hierarchies in topic models. Experimental results with four baselines show that HyHTM can better attend to parent-child relationships among topics. HyHTM produces coherent topic hierarchies that specialise in granularity from generic higher-level topics to specific lowerlevel topics. Further, our model is significantly faster and leaves a much smaller memory footprint than our best-performing baseline.We have made the source code for our algorithm publicly accessible.
△ Less
Submitted 16 May, 2023;
originally announced May 2023.
-
Partial-input baselines show that NLI models can ignore context, but they don't
Authors:
Neha Srikanth,
Rachel Rudinger
Abstract:
When strong partial-input baselines reveal artifacts in crowdsourced NLI datasets, the performance of full-input models trained on such datasets is often dismissed as reliance on spurious correlations. We investigate whether state-of-the-art NLI models are capable of overriding default inferences made by a partial-input baseline. We introduce an evaluation set of 600 examples consisting of perturb…
▽ More
When strong partial-input baselines reveal artifacts in crowdsourced NLI datasets, the performance of full-input models trained on such datasets is often dismissed as reliance on spurious correlations. We investigate whether state-of-the-art NLI models are capable of overriding default inferences made by a partial-input baseline. We introduce an evaluation set of 600 examples consisting of perturbed premises to examine a RoBERTa model's sensitivity to edited contexts. Our results indicate that NLI models are still capable of learning to condition on context--a necessary component of inferential reasoning--despite being trained on artifact-ridden datasets.
△ Less
Submitted 24 May, 2022;
originally announced May 2022.
-
Elaborative Simplification: Content Addition and Explanation Generation in Text Simplification
Authors:
Neha Srikanth,
Junyi Jessy Li
Abstract:
Much of modern-day text simplification research focuses on sentence-level simplification, transforming original, more complex sentences into simplified versions. However, adding content can often be useful when difficult concepts and reasoning need to be explained. In this work, we present the first data-driven study of content addition in text simplification, which we call elaborative simplificat…
▽ More
Much of modern-day text simplification research focuses on sentence-level simplification, transforming original, more complex sentences into simplified versions. However, adding content can often be useful when difficult concepts and reasoning need to be explained. In this work, we present the first data-driven study of content addition in text simplification, which we call elaborative simplification. We introduce a new annotated dataset of 1.3K instances of elaborative simplification in the Newsela corpus, and analyze how entities, ideas, and concepts are elaborated through the lens of contextual specificity. We establish baselines for elaboration generation using large-scale pre-trained language models, and demonstrate that considering contextual specificity during generation can improve performance. Our results illustrate the complexities of elaborative simplification, suggesting many interesting directions for future work.
△ Less
Submitted 3 June, 2021; v1 submitted 20 October, 2020;
originally announced October 2020.
-
Mass transfer in Frenkel-Kontorova chain initiated by molecule impact
Authors:
A. Moradi Marjaneh,
D. Saadatmand,
I. Evazzade,
R. I. Babicheva,
E. G. Soboleva,
N. Srikanth,
Kun Zhou,
E. A. Korznikova,
S. V. Dmitriev
Abstract:
The Frenkel-Kontorova chain with a free end is used to study initiation and propagation of crowdions (anti-kinks) caused by impact of a molecule consisting of K atoms. It is found that molecules with 1 < K < 10 are more efficient in initiation of crowdions as compared to single atom (K = 1) because total energy needed to initiate the crowdions by molecules is smaller. This happens because single a…
▽ More
The Frenkel-Kontorova chain with a free end is used to study initiation and propagation of crowdions (anti-kinks) caused by impact of a molecule consisting of K atoms. It is found that molecules with 1 < K < 10 are more efficient in initiation of crowdions as compared to single atom (K = 1) because total energy needed to initiate the crowdions by molecules is smaller. This happens because single atom can initiate in the chain only sharp, fast-moving crowdions that requires a relatively large energy. Molecule has finite length, that is why it is able to excite a wider crowdion with a smaller velocity and smaller energy. Our results can shed light on the atomistic mechanisms of mass transfer in crystals subject to atom and molecule bombardment.
△ Less
Submitted 18 May, 2018;
originally announced May 2018.
-
Exploring the Charge Localization and Band Gap Opening of Borophene: A First-Principles Study
Authors:
Andrey A. Kistanov,
Yongqing Cai,
Kun Zhou,
Narasimalu Srikanth,
Sergey V. Dmitriev,
Yong-Wei Zhang
Abstract:
Recently synthesized two-dimensional (2D) boron, borophene, exhibits a novel metallic behavior rooted in the s-p orbital hybridization, distinctively different from other 2D materials such as sulfides/selenides and semi-metallic graphene. This unique feature of borophene implies new routes for charge delocalization and band gap opening. Herein, using first-principles calculations, we explore the r…
▽ More
Recently synthesized two-dimensional (2D) boron, borophene, exhibits a novel metallic behavior rooted in the s-p orbital hybridization, distinctively different from other 2D materials such as sulfides/selenides and semi-metallic graphene. This unique feature of borophene implies new routes for charge delocalization and band gap opening. Herein, using first-principles calculations, we explore the routes to localize the carriers and open the band gap of borophene via chemical functionalization, ribbon construction, and defect engineering. The metallicity of borophene is found to be remarkably robust against H- and F-functionalization and the presence of vacancies. Interestingly, a strong odd-even oscillation of the electronic structure with width is revealed for H-functionalized borophene nanoribbons, while an ultra-high work function (~ 7.83 eV) is found for the F-functionalized borophene due to its strong charge transfer to the atomic adsorbates.
△ Less
Submitted 13 December, 2017;
originally announced December 2017.
-
On a problem of resonance with exponential non linearity
Authors:
B. B. Manna,
P. N. Srikanth
Abstract:
We have considered the following semi linear elliptic problem on the unit disk $B$ $-Δu = λ_1 u+e^u+f $ in $B$ with the Dirichlet boundary condition and $f$ satisfying the following condition : $f\in L^r(B)$, for some $r>2$ and $-\int_B fφ_1<4π$. Where $φ_1$ is the eigen function of $(-Δ)$ corresponding to the first eigenvalue $λ_1$ in $H_0^1(B)$. We shall find the existence of a radial solution o…
▽ More
We have considered the following semi linear elliptic problem on the unit disk $B$ $-Δu = λ_1 u+e^u+f $ in $B$ with the Dirichlet boundary condition and $f$ satisfying the following condition : $f\in L^r(B)$, for some $r>2$ and $-\int_B fφ_1<4π$. Where $φ_1$ is the eigen function of $(-Δ)$ corresponding to the first eigenvalue $λ_1$ in $H_0^1(B)$. We shall find the existence of a radial solution of this PDE. We shall use degree theory to get the existence starting from a suitable with known solution with its degree. Connecting those two PDE's by homotopy and getting the uniform estimate for the connecting PDE's we shall achieve our result.
△ Less
Submitted 9 May, 2016; v1 submitted 5 May, 2014;
originally announced May 2014.
-
On the solutions of a singular elliptic equation concentrating on a circle
Authors:
B. B. Manna,
P. N. Srikanth
Abstract:
Let $A=\{x\in \R^{2N+2} : 0< a< |x| <b\}$ be an annulus. Consider the following singularly perturbed elliptic problem on $A$
\begin{equation}
\begin{array}{lll}
-\eps^2{\De u} + |x|^αu = |x|^αu^p, &\mbox{\qquad in} A \notag u>0 &\mbox{\qquad in} A
\frac{\partial u}{\partialν} = 0 &\mbox{\qquad on} \partial A
\end{array} %\label{a1}
\end{equation} $1<p<2^*-1$. We shall show that there e…
▽ More
Let $A=\{x\in \R^{2N+2} : 0< a< |x| <b\}$ be an annulus. Consider the following singularly perturbed elliptic problem on $A$
\begin{equation}
\begin{array}{lll}
-\eps^2{\De u} + |x|^αu = |x|^αu^p, &\mbox{\qquad in} A \notag u>0 &\mbox{\qquad in} A
\frac{\partial u}{\partialν} = 0 &\mbox{\qquad on} \partial A
\end{array} %\label{a1}
\end{equation} $1<p<2^*-1$. We shall show that there exists a positive solution $u_\eps$ concentrating on an $S^1$ orbit as $\eps\to 0$. We prove this by reducing the problem to a lower dimensional one and analyzing a single point concentrating solution in the lower dimensional space. We make precise how the single peak concentration depends on the parameter $α$.
△ Less
Submitted 22 October, 2013;
originally announced October 2013.
-
On the solutions of a singular elliptic equation concentrating on two orthogonal spheres
Authors:
B. B. Manna,
P. N. Srikanth
Abstract:
Let $A=\{x\in \R^{2m} : 0< a< |x| <b\}$ be an annulus. Consider the following singularly perturbed elliptic problem on $A$
\begin{equation}
\begin{array}{lll}
-\eps^2{\De u} + |x|^ηu = |x|^ηu^p, &\mbox{\qquad in} A \notag u>0 &\mbox{\qquad in} A u = 0 &\mbox{\qquad on} \partial A
\end{array} %\label{a1}
\end{equation} $1<p<2^*-1$. We shall prove the existence of a positive solution…
▽ More
Let $A=\{x\in \R^{2m} : 0< a< |x| <b\}$ be an annulus. Consider the following singularly perturbed elliptic problem on $A$
\begin{equation}
\begin{array}{lll}
-\eps^2{\De u} + |x|^ηu = |x|^ηu^p, &\mbox{\qquad in} A \notag u>0 &\mbox{\qquad in} A u = 0 &\mbox{\qquad on} \partial A
\end{array} %\label{a1}
\end{equation} $1<p<2^*-1$. We shall prove the existence of a positive solution $u_\eps$ which concentrates on two different orthogonal spheres of dimension $(m-1)$ as $\eps\to 0$. We achieve this by studying a reduced problem on an annular domain in $\R^{m+1}$ and analyzing the profile of a two point concentrating solution in this domain.
△ Less
Submitted 23 September, 2013;
originally announced September 2013.
-
A Reduction Method for Semilinear Elliptic Equations and Solutions Concentrating on Spheres
Authors:
Filomena Pacella,
P. N. Srikanth
Abstract:
We show that any general semilinear elliptic problem with Dirichlet or Neumann boundary conditions in an annulus A in R^2m ;m >1, invariant by the action of a certain symmetry group can be reduced to a nonhomogenous similar problem in an annulus D in R^(m+1), invariant by another related symmetry. We apply this result to prove the existence of positive and sign changing solutions of a singularly p…
▽ More
We show that any general semilinear elliptic problem with Dirichlet or Neumann boundary conditions in an annulus A in R^2m ;m >1, invariant by the action of a certain symmetry group can be reduced to a nonhomogenous similar problem in an annulus D in R^(m+1), invariant by another related symmetry. We apply this result to prove the existence of positive and sign changing solutions of a singularly perturbed elliptic problem in A which concentrate on one or two (m-1) dimensional spheres. We also prove that the Morse indices of these solutions tend to infinity as the parameter of concentration tends to infinity.
△ Less
Submitted 2 October, 2012;
originally announced October 2012.
-
Survey on Various Gesture Recognition Techniques for Interfacing Machines Based on Ambient Intelligence
Authors:
Harshith C,
Karthik R. Shastry,
Manoj Ravindran,
M. V. V. N. S. Srikanth,
Naveen Lakshmikhanth
Abstract:
Gesture recognition is mainly apprehensive on analyzing the functionality of human wits. The main goal of gesture recognition is to create a system which can recognize specific human gestures and use them to convey information or for device control. Hand gestures provide a separate complementary modality to speech for expressing ones ideas. Information associated with hand gestures in a conversati…
▽ More
Gesture recognition is mainly apprehensive on analyzing the functionality of human wits. The main goal of gesture recognition is to create a system which can recognize specific human gestures and use them to convey information or for device control. Hand gestures provide a separate complementary modality to speech for expressing ones ideas. Information associated with hand gestures in a conversation is degree,discourse structure, spatial and temporal structure. The approaches present can be mainly divided into Data-Glove Based and Vision Based approaches. An important face feature point is the nose tip. Since nose is the highest protruding point from the face. Besides that, it is not affected by facial expressions.Another important function of the nose is that it is able to indicate the head pose. Knowledge of the nose location will enable us to align an unknown 3D face with those in a face database. Eye detection is divided into eye position detection and eye contour detection. Existing works in eye detection can be classified into two major categories: traditional image-based passive approaches and the active IR based approaches. The former uses intensity and shape of eyes for detection and the latter works on the assumption that eyes have a reflection under near IR illumination and produce bright/dark pupil effect. The traditional methods can be broadly classified into three categories: template based methods,appearance based methods and feature based methods. The purpose of this paper is to compare various human Gesture recognition systems for interfacing machines directly to human wits without any corporeal media in an ambient environment.
△ Less
Submitted 30 November, 2010;
originally announced December 2010.
