CoSy: Conversational Synthesis for Grounded Question Answering

Leveraging Large Language Models to Detect and Revise Unsafe Responses in Context-Sensitive Dialogues

Large Language Models (LLMs) excel at tasks like classification, summarisation, question answering among others, with performance comparable to humans. Despite these capabilities, leveraging LLMs to transform unsafe responses in context-sensitive dialogues is underexplored. In this work, we propose a pipeline that leverage LLMs as safety detector, editor and evaluator to mitigate undesired behaviour in human-computer dialogues. At the first iteration, our experimental results on two evaluation datasets show reduction in the unsafe dialogues from 47% to 13% and 48% to 2% respectively, with 82% and 92% agreement between the safety detector and evaluator after revision. Human evaluation of randomly sampled dialogues demonstrates reduction in unsafe responses after revision. Additionally, the revision LLM (editor) exhibits a higher proportion of refusals without compromising fluency and coherence of the revised dialogues.

Toward Exploring Mixed-Initiative Conversation Generation Based on Community Question Answering

Conversational search addresses users’ information needs through multi-turn and context-aware interactions. Given that user queries are often ambiguous, the use of clarifying questions can effectively reduce uncertainty and enable a mixed-initiative conversational system. However, current datasets for clarifying questions remain limited in the following three aspects: (1) underrepresented multi-turn conversational data, (2) limited diversity, and (3) heavily reliance on crowdsourcing, thereby suffering from limitations such as high annotation cost. To address these issues, we propose a large language model (LLM)-based three-stage framework that relies on an existing community question answering dataset. It encompasses: (1) extracting essential information from the initial user query with the relevant contextual information, (2) generating clarifying questions paired with corresponding answers, and (3) refining conversations to ensure coherence and a natural conversational flow. We assess our multi-stage method against a baseline that directly prompts LLMs to generate conversations in a single-step process, evaluating on an answer retrieval task using recall, precision, normalized discounted cumulative gain and mean average precision. Results show that our three-stage generation approach consistently outperforms the baseline particularly in recall, while also achieving competitive results across other metrics. Human and automatic evaluations further indicate the high quality of generated conversations and fine-tuning on them improves retrieval performance, highlighting the pipeline’s potential.

TurnBench: A Multi-Domain Benchmark for Turn-Taking Dynamics in Spoken Dialogue

Speakers in natural conversation take turns speaking and listening, deciding in real time when to take, hold, or yield the floor. However, turn-taking evaluation remains limited due to the lack of a consistent, linguistically grounded evaluation protocol and hand-annotated data covering diverse conversation types. To address this, we present TurnBench, a multi-domain benchmark that pairs a 30-hour, hand-labeled corpus of dyadic human conversation with a standardized evaluation protocol for end-of-turn and interruption detection. We set conversation type as a controllable experimental variable, covering six distinct interaction styles, and triple-annotate each conversation. Benchmarking 14 heterogeneous turn-taking systems, we find end-of-turn recall stable across types, while interruption false positives are strongly type-dependent and concentrated in backchannel-dense interaction styles. Although in smooth floor transfers human listeners begin speaking a median 151 ms before the current turn ends, no current system performs equivalently without incurring excessive false positives. We release our corpus, a 104-hour training set, and a public leaderboard with an interactive dataset viewer at https://turnbench.sesame.com

Conversational Grounding in Large Language Models: Evaluation Methods, Challenges and Future Directions

LLM performance on multi-interlocutor NLI tasks

KGA-LM: Representation-Level Grounding for Conversational Search over Knowledge Graphs

Integrating structured knowledge graphs (KGs) with Large Language Models (LLMs) is essential for trustworthy, knowledge intensive conversational systems. However, existing Retrieval Augmented Generation (RAG) methods typically rely on a retrieval-as-context paradigm that linearizes structured subgraphs into unstructured prompt tokens. This approach not only flattens rich structural dependencies but also leads to context inflation and evidence attenuation in multi-turn dialogues. To address these limitations, we propose KGA-LM, a framework that integrates external knowledge via representation-level grounding. Rather than treating retrieved evidence as transient input artifacts, KGA-LM encodes compact multi-hop subgraphs using a Graph Transformer and fuses them into the LLM decoder through a compatibility-aware latent interface. This design aligns the heterogeneous latent spaces of the graph encoder and the LLM, while a dual-gated fusion mechanism dynamically regulates the influence of non-parametric graph evidence across turns. Experiments on multiple conversational benchmarks demonstrate that KGA-LM significantly improves factual accuracy and reduces hallucination compared to prompt-linearized baselines. Crucially, by decoupling knowledge injection from prompt length, our approach mitigates retrieval signal decay under long contexts, offering a superior trade-off between grounding quality and inference efficiency.