Mohsen Mesgar

Haau-Sing Li, Mohsen Mesgar, André F. T. Martins et al.

Code generation from text requires understanding the user's intent from a natural language description and generating an executable code snippet that satisfies this intent. While recent pretrained language models demonstrate remarkable performance for this task, these models fail when the given natural language description is under-specified. In this work, we introduce a novel and more realistic setup for this task. We hypothesize that the under-specification of a natural language description can be resolved by asking clarification questions. Therefore, we collect and introduce a new dataset named CodeClarQA containing pairs of natural language descriptions and code with created synthetic clarification questions and answers. The empirical results of our evaluation of pretrained language model performance on code generation show that clarifications result in more precisely generated code, as shown by the substantial improvement of model performance in all evaluation metrics. Alongside this, our task and dataset introduce new challenges to the community, including when and what clarification questions should be asked. Our code and dataset are available on GitHub.

Mohsen Mesgar, Thy Thy Tran, Goran Glavas et al.

Few-shot Intent Classification (FSIC) is one of the key challenges in modular task-oriented dialog systems. While advanced FSIC methods are similar in using pretrained language models to encode texts and nearest neighbour-based inference for classification, these methods differ in details. They start from different pretrained text encoders, use different encoding architectures with varying similarity functions, and adopt different training regimes. Coupling these mostly independent design decisions and the lack of accompanying ablation studies are big obstacle to identify the factors that drive the reported FSIC performance. We study these details across three key dimensions: (1) Encoding architectures: Cross-Encoder vs Bi-Encoders; (2) Similarity function: Parameterized (i.e., trainable) functions vs non-parameterized function; (3) Training regimes: Episodic meta-learning vs the straightforward (i.e., non-episodic) training. Our experimental results on seven FSIC benchmarks reveal three important findings. First, the unexplored combination of the cross-encoder architecture (with parameterized similarity scoring function) and episodic meta-learning consistently yields the best FSIC performance. Second, Episodic training yields a more robust FSIC classifier than non-episodic one. Third, in meta-learning methods, splitting an episode to support and query sets is not a must. Our findings paves the way for conducting state-of-the-art research in FSIC and more importantly raise the community's attention to details of FSIC methods. We release our code and data publicly.

Pascal Tilli, Mohsen Mesgar

Visual Document Retrieval (VDR) models mostly rely on late interaction architectures, in which documents are represented by a set of local patch embeddings and then matched against query tokens. While efficient, this architecture prioritizes local similarity over global layout structure of documents to estimate relevancy between documents and query. In practice, this leads to errors as relevance originates from layout structure of documents with heterogeneous layouts combining figures, tables, and text. We make document layout learnable without changing inference. We propose a multimodal encoder that augments local patch representations with a global layout embedding, trained via textual descriptions encoding document layout information. Across four ViDoRe-v2 datasets, our model improves over the strongest architecturally comparable ColPali/ColQwen baseline by +2.4 nDCG@5 and +2.3 MAP@5, with statistically significant per-dataset gains over ColQwen.

Wei Zhou, Mohsen Mesgar, Heike Adel et al.

Table Question Answering (TQA) aims at composing an answer to a question based on tabular data. While prior research has shown that TQA models lack robustness, understanding the underlying cause and nature of this issue remains predominantly unclear, posing a significant obstacle to the development of robust TQA systems. In this paper, we formalize three major desiderata for a fine-grained evaluation of robustness of TQA systems. They should (i) answer questions regardless of alterations in table structure, (ii) base their responses on the content of relevant cells rather than on biases, and (iii) demonstrate robust numerical reasoning capabilities. To investigate these aspects, we create and publish a novel TQA evaluation benchmark in English. Our extensive experimental analysis reveals that none of the examined state-of-the-art TQA systems consistently excels in these three aspects. Our benchmark is a crucial instrument for monitoring the behavior of TQA systems and paves the way for the development of robust TQA systems. We release our benchmark publicly.

Wei Zhou, Mohsen Mesgar, Annemarie Friedrich et al.

Complex table question answering (TQA) aims to answer questions that require complex reasoning, such as multi-step or multi-category reasoning, over data represented in tabular form. Previous approaches demonstrated notable performance by leveraging either closed-source large language models (LLMs) or fine-tuned open-weight LLMs. However, fine-tuning LLMs requires high-quality training data, which is costly to obtain, and utilizing closed-source LLMs poses accessibility challenges and leads to reproducibility issues. In this paper, we propose Multi-Agent Collaboration with Tool use (MACT), a framework that requires neither closed-source models nor fine-tuning. In MACT, a planning agent and a coding agent that also make use of tools collaborate to answer questions. Our experiments on four TQA benchmarks show that MACT outperforms previous SoTA systems on three out of four benchmarks and that it performs comparably to the larger and more expensive closed-source model GPT-4 on two benchmarks, even when using only open-weight models without any fine-tuning. We conduct extensive analyses to prove the effectiveness of MACT's multi-agent collaboration in TQA.

Wei Zhou, Mohsen Mesgar, Heike Adel et al.

In table question answering (TQA), tables are encoded as either texts or images. Prior work suggests that passing images of tables to multi-modal large language models (MLLMs) performs comparably to or even better than using textual input with large language models (LLMs). However, the lack of controlled setups limits fine-grained distinctions between these approaches. In this paper, we conduct the first controlled study on the effectiveness of several combinations of table representations and models from two perspectives: question complexity and table size. We build a new benchmark based on existing TQA datasets. In a systematic analysis of seven pairs of MLLMs and LLMs, we find that the best combination of table representation and model varies across setups. We propose FRES, a method selecting table representations dynamically, and observe a 10% average performance improvement compared to using both representations indiscriminately.

Lei Tang, Wei Zhou, Mohsen Mesgar

Process reward models (PRMs) improve complex reasoning in large language models (LLMs) by grading candidate solutions step-by-step and selecting answers via aggregated step scores. While effective in domains such as mathematics, their applicability to tasks involving semi-structured data, like table question answering (TQA) remains unexplored. TQA poses unique challenges for PRMs, including abundant irrelevant information, loosely connected reasoning steps, and domain-specific reasoning. This work presents the first systematic study of PRMs for TQA. We evaluate state-of-the-art generative PRMs on TQA from both answer and step perspectives. Results show that PRMs that combine textual and code verification can aid solution selection but struggle to generalize to out-of-domain data. Analysis reveals a weak correlation between performance in step-level verification and answer accuracy, possibly stemming from weak step dependencies and loose causal links. Our findings highlight limitations of current PRMs on TQA and offer valuable insights for building more robust, process-aware verifiers.

Wei Zhou, Bolei Ma, Annemarie Friedrich et al.

Table Question Answering (TQA) aims to answer natural language questions about tabular data, often accompanied by additional contexts such as text passages. The task spans diverse settings, varying in table representation, question/answer complexity, modality involved, and domain. While recent advances in large language models (LLMs) have led to substantial progress in TQA, the field still lacks a systematic organization and understanding of task formulations, core challenges, and methodological trends, particularly in light of emerging research directions such as reinforcement learning. This survey addresses this gap by providing a comprehensive and structured overview of TQA research with a focus on LLM-based methods. We provide a comprehensive categorization of existing benchmarks and task setups. We group current modeling strategies according to the challenges they target, and analyze their strengths and limitations. Furthermore, we highlight underexplored but timely topics that have not been systematically covered in prior research. By unifying disparate research threads and identifying open problems, our survey offers a consolidated foundation for the TQA community, enabling a deeper understanding of the state of the art and guiding future developments in this rapidly evolving area.

Wei Zhou, Mohsen Mesgar, Heike Adel et al.

Table question answering (TQA) focuses on answering questions based on tabular data. Developing TQA systems targets effective interaction with tabular data for tasks such as cell retrieval and data analysis. While recent work has leveraged fine-tuning to improve TQA systems, existing approaches often under-utilize available data and neglect the potential of post-training for further gains. In this work, we introduce p2-TQA, a process-based preference learning framework for TQA post-training. p2-TQA automatically constructs process-based preference data via a table-specific pipeline, eliminating the need for manual or costly data collection. It then optimizes models through contrastive learning on the collected data. Experiments show that p2-TQA effectively improves TQA models by up to 5% on in-domain datasets and 2.4% on out-of-domain datasets with only 8,000 training instances. Furthermore, models enhanced with p2-TQA achieve competitive results against larger, more complex state-of-the-art TQA systems, while maintaining up to five times higher efficiency.

Wen Lai, Mohsen Mesgar, Alexander Fraser

To democratize large language models (LLMs) to most natural languages, it is imperative to make these models capable of understanding and generating texts in many languages, in particular low-resource ones. While recent multilingual LLMs demonstrate remarkable performance in such capabilities, these LLMs still support a limited number of human languages due to the lack of training data for low-resource languages. Moreover, these LLMs are not yet aligned with human preference for downstream tasks, which is crucial for the success of LLMs in English. In this paper, we introduce xLLaMA-100 and xBLOOM-100 (collectively xLLMs-100), which scale the multilingual capabilities of LLaMA and BLOOM to 100 languages. To do so, we construct two datasets: a multilingual instruction dataset including 100 languages, which represents the largest language coverage to date, and a cross-lingual human feedback dataset encompassing 30 languages. We perform multilingual instruction tuning on the constructed instruction data and further align the LLMs with human feedback using the DPO algorithm on our cross-lingual human feedback dataset. We evaluate the multilingual understanding and generating capabilities of xLLMs-100 on five multilingual benchmarks. Experimental results show that xLLMs-100 consistently outperforms its peers across the benchmarks by considerable margins, defining a new state-of-the-art multilingual LLM that supports 100 languages.

Federico Ruggeri, Mohsen Mesgar, Iryna Gurevych

The applications of conversational agents for scientific disciplines (as expert domains) are understudied due to the lack of dialogue data to train such agents. While most data collection frameworks, such as Amazon Mechanical Turk, foster data collection for generic domains by connecting crowd workers and task designers, these frameworks are not much optimized for data collection in expert domains. Scientists are rarely present in these frameworks due to their limited time budget. Therefore, we introduce a novel framework to collect dialogues between scientists as domain experts on scientific papers. Our framework lets scientists present their scientific papers as groundings for dialogues and participate in dialogue they like its paper title. We use our framework to collect a novel argumentative dialogue dataset, ArgSciChat. It consists of 498 messages collected from 41 dialogues on 20 scientific papers. Alongside extensive analysis on ArgSciChat, we evaluate a recent conversational agent on our dataset. Experimental results show that this agent poorly performs on ArgSciChat, motivating further research on argumentative scientific agents. We release our framework and the dataset.

Mohsen Mesgar, Edwin Simpson, Iryna Gurevych

Neural models for response generation produce responses that are semantically plausible but not necessarily factually consistent with facts describing the speaker's persona. These models are trained with fully supervised learning where the objective function barely captures factual consistency. We propose to fine-tune these models by reinforcement learning and an efficient reward function that explicitly captures the consistency between a response and persona facts as well as semantic plausibility. Our automatic and human evaluations on the PersonaChat corpus confirm that our approach increases the rate of responses that are factually consistent with persona facts over its supervised counterpart while retaining the language quality of responses.

Mohsen Mesgar, Paul Youssef, Lin Li et al.

Our conversational agent UKP-ATHENA assists NLP researchers in finding and exploring scientific literature, identifying relevant authors, planning or post-processing conference visits, and preparing paper submissions using a unified interface based on natural language inputs and responses. UKP-ATHENA enables new access paths to our swiftly evolving research area with its massive amounts of scientific information and high turnaround times. UKP-ATHENA's responses connect information from multiple heterogeneous sources which researchers currently have to explore manually one after another. Unlike a search engine, UKP-ATHENA maintains the context of a conversation to allow for efficient information access on papers, researchers, and conferences. Our architecture consists of multiple components with reference implementations that can be easily extended by new skills and domains. Our user-based evaluation shows that UKP-ATHENA already responds 45% of different formulations of defined intents with 37% information coverage rate.

Mohsen Mesgar, Sebastian Bücker, Iryna Gurevych

Recent dialogue coherence models use the coherence features designed for monologue texts, e.g. nominal entities, to represent utterances and then explicitly augment them with dialogue-relevant features, e.g., dialogue act labels. It indicates two drawbacks, (a) semantics of utterances is limited to entity mentions, and (b) the performance of coherence models strongly relies on the quality of the input dialogue act labels. We address these issues by introducing a novel approach to dialogue coherence assessment. We use dialogue act prediction as an auxiliary task in a multi-task learning scenario to obtain informative utterance representations for coherence assessment. Our approach alleviates the need for explicit dialogue act labels during evaluation. The results of our experiments show that our model substantially (more than 20 accuracy points) outperforms its strong competitors on the DailyDialogue corpus, and performs on par with them on the SwitchBoard corpus for ranking dialogues concerning their coherence.

Yang Gao, Christian M. Meyer, Mohsen Mesgar et al.

Document summarisation can be formulated as a sequential decision-making problem, which can be solved by Reinforcement Learning (RL) algorithms. The predominant RL paradigm for summarisation learns a cross-input policy, which requires considerable time, data and parameter tuning due to the huge search spaces and the delayed rewards. Learning input-specific RL policies is a more efficient alternative but so far depends on handcrafted rewards, which are difficult to design and yield poor performance. We propose RELIS, a novel RL paradigm that learns a reward function with Learning-to-Rank (L2R) algorithms at training time and uses this reward function to train an input-specific RL policy at test time. We prove that RELIS guarantees to generate near-optimal summaries with appropriate L2R and RL algorithms. Empirically, we evaluate our approach on extractive multi-document summarisation. We show that RELIS reduces the training time by two orders of magnitude compared to the state-of-the-art models while performing on par with them.

Steffen Eger, Gözde Gül Şahin, Andreas Rücklé et al.

Visual modifications to text are often used to obfuscate offensive comments in social media (e.g., "!d10t") or as a writing style ("1337" in "leet speak"), among other scenarios. We consider this as a new type of adversarial attack in NLP, a setting to which humans are very robust, as our experiments with both simple and more difficult visual input perturbations demonstrate. We then investigate the impact of visual adversarial attacks on current NLP systems on character-, word-, and sentence-level tasks, showing that both neural and non-neural models are, in contrast to humans, extremely sensitive to such attacks, suffering performance decreases of up to 82\%. We then explore three shielding methods---visual character embeddings, adversarial training, and rule-based recovery---which substantially improve the robustness of the models. However, the shielding methods still fall behind performances achieved in non-attack scenarios, which demonstrates the difficulty of dealing with visual attacks.