Deep research has emerged as an important task that aims to address hard queries that need extensive open-web exploration. To tackle it, most prior work equips large language model (LLM)-based agents with opaque web search APIs, enabling agents to iteratively issue search queries, retrieve external evidence, and reason over it. Despite search's essential role in deep research, black-box web search APIs leave the behaviour of established text ranking methods in deep research largely unclear. To fill this gap, we reproduce key findings and best practices for text ranking methods in deep research. We examine their effectiveness from three perspectives: (i) retrieval units (documents vs. passages), (ii) pipeline configurations (different retrievers, re-rankers, and re-ranking depths), and (iii) query characteristics (the mismatch between agent-issued queries and the training queries of text rankers). We perform experiments on BrowseComp-Plus, a deep research dataset with a fixed corpus, evaluating 2 open-source agents, 5 retrievers, and 3 re-rankers. We find that agent-issued queries typically follow web-search-style syntax (e.g., quoted exact matches), favouring lexical, learned sparse, and multi-vector retrievers; passage-level units are more efficient under limited context windows, and avoid the difficulties of document length normalisation in lexical retrieval; re-ranking is highly effective. We further propose a query-to-question (Q2Q) method that translates agent-issued queries into natural-language questions, significantly reducing the query mismatch.

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.

Fengran Mo, Yifan Gao, Chuan Meng, Xin Liu, Zhuofeng Wu, Kelong Mao, Zhengyang Wang, Pei Chen, Zheng Li, Xian Li, Bing Yin, Meng Jiang. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Proactive search in conversations (PSC) aims to reduce user effort in formulating explicit queries by proactively retrieving useful relevant information given conversational context. Previous work in PSC either directly uses this context as input to off-the-shelf ad-hoc retrievers or further fine-tunes them on PSC data. However, ad-hoc retrievers are pre-trained on short and concise queries, while the PSC input is longer and noisier. This input mismatch between ad-hoc search and PSC limits retrieval quality. While fine-tuning on PSC data helps, its benefits remain constrained by this input gap. In this work, we propose Conv2Query, a novel conversation-to-query framework that adapts ad-hoc retrievers to PSC by bridging the input gap between ad-hoc search and PSC. Conv2Query maps conversational context into ad-hoc queries, which can either be used as input for off-the-shelf ad-hoc retrievers or for further fine-tuning on PSC data. Extensive experiments on two PSC datasets show that Conv2Query significantly improves ad-hoc retrievers' performance, both when used directly and after fine-tuning on PSC.

Conversational search enables multi-turn interactions between users and systems to fulfill users' complex information needs. During this interaction, the system should understand the users' search intent within the conversational context and then return the relevant information through a flexible, dialogue-based interface. The recent powerful large language models (LLMs) with capacities of instruction following, content generation, and reasoning, attract significant attention and advancements, providing new opportunities and challenges for building up intelligent conversational search systems. This tutorial aims to introduce the connection between fundamentals and the emerging topics revolutionized by LLMs in the context of conversational search. It is designed for students, researchers, and practitioners from both academia and industry. Participants will gain a comprehensive understanding of both the core principles and cutting-edge developments driven by LLMs in conversational search, equipping them with the knowledge needed to contribute to the development of next-generation conversational search systems.

Query performance prediction (QPP) is a key task in information retrieval (IR), focusing on estimating the retrieval quality of a given query without relying on human-labeled relevance judgments. Over the decades, QPP has gained increasing significance, with a surge in research activity in recent years. It has proven to benefit various aspects of retrieval, such as optimizing retrieval effectiveness by selecting the most appropriate ranking function for each query.

Query performance prediction (QPP) aims to estimate the retrieval quality of a search system for a query without human relevance judgments. Previous QPP methods typically return a single scalar value and do not require the predicted values to approximate a specific information retrieval (IR) evaluation measure, leading to certain drawbacks: (i) a single scalar is insufficient to accurately represent different IR evaluation measures, especially when metrics do not highly correlate, and (ii) a single scalar limits the interpretability of QPP methods because solely using a scalar is insufficient to explain QPP results. To address these issues, we propose a QPP framework using automatically gen erated re levance judgments (QPP-GenRE), which decomposes QPP into independent subtasks of predicting the relevance of each item in a ranked list to a given query. This allows us to predict any IR evaluation measure using the generated relevance judgments as pseudo-labels. This also allows us to interpret predicted IR evaluation measures, and identify, track, and rectify errors in generated relevance judgments to improve QPP quality. We predict an item’s relevance by using open source large language models (LLMs) to ensure scientific reproducibility. We face two main challenges: (i) excessive computational costs of judging an entire corpus for predicting a metric considering recall, and (ii) limited performance in prompting open source LLMs in a zero-/few-shot manner. To solve the challenges, we devise an approximation strategy to predict an IR measure considering recall and propose to fine-tune open source LLMs using human-labeled relevance judgments. Experiments on the TREC 2019–2022 deep learning tracks and CAsT-19–20 datasets show that QPP-GenRE achieves state-of-the-art QPP quality for both lexical and neural rankers.

Query performance prediction (QPP) is a key task in information retrieval (IR) [1]. The QPP task is to estimate the retrieval quality of a search system for a query without human relevance judgments. In summary, I aim to solve 4 limitations identified in previous QPP studies: I have published 3 papers that address 3 of these limitations, while the remaining one is the focus of my future work.

We study ranked list truncation (RLT) from a novel retrieve-then-re-rank perspective, where we optimize re-ranking by truncating the retrieved list (i.e., trim re-ranking candidates). RLT is crucial for re-ranking as it can improve re-ranking efficiency by sending variable-length candidate lists to a re-ranker on a per-query basis. It also has the potential to improve re-ranking effectiveness. Despite its importance, there is limited research into applying RLT methods to this new perspective. To address this research gap, we reproduce existing RLT methods in the context of re-ranking, especially newly emerged large language model (LLM)-based re-ranking. In particular, we examine to what extent established findings on RLT for retrieval are generalizable to the "retrieve-then-re-rank" setup from three perspectives: (i) assessing RLT methods in the context of LLM-based re-ranking with lexical first-stage retrieval, (ii) investigating the impact of different types of first-stage retrievers on RLT methods, and (iii) investigating the impact of different types of re-rankers on RLT methods. We perform experiments on the TREC 2019 and 2020 deep learning tracks, investigating 8 RLT methods for pipelines involving 3 retrievers and 2 re-rankers. We reach new insights into RLT methods in the context of re-ranking.

Identifying the right moment for a system to take the initiative is essential to conversational information seeking (CIS). Existing studies have extensively studied the clarification need prediction task, i.e., predicting when to ask a clarifying question, however, it only covers one specific system-initiative action. We define the system initiative prediction (SIP) task as predicting whether a CIS system should take the initiative at the next turn. Our analysis reveals that for effective modeling of SIP, it is crucial to capture dependencies between adjacent user?system initiative-taking decisions. We propose to model SIP by CRFs. Due to their graphical nature, CRFs are effective in capturing such dependencies and have greater transparency than more complex methods, e.g., LLMs. Applying CRFs to SIP comes with two challenges: (i) CRFs need to be given the unobservable system utterance at the next turn, and (ii) they do not explicitly model multi-turn features. We model SIP as an input-incomplete sequence labeling problem and propose a multi-turn system initiative predictor (MuSIc) that has (i) prior-posterior inter-utterance encoders to eliminate the need to be given the unobservable system utterance, and (ii) a multi-turn feature-aware CRF layer to incorporate multi-turn features into the dependencies between adjacent initiative-taking decisions. Experiments show that MuSIc outperforms LLM-based baselines including LLaMA, achieving state-of-the-art results on SIP. We also show the benefits of SIP on clarification need prediction and action prediction.

Generating synthetic training data based on large language models (LLMs) for ranking models has gained attention recently.Prior studies use LLMs to build pseudo querydocument pairs by generating synthetic queries from documents in a corpus.In this paper, we propose a new perspective of data augmentation: generating synthetic documents from queries.To achieve this, we propose DocGen, that consists of a three-step pipeline that utilizes the few-shot capabilities of LLMs.Doc-Gen pipeline performs synthetic document generation by (i) expanding, (ii) highlighting the original query, and then (iii) generating a synthetic document that is likely to be relevant to the query.To further improve the relevance between generated synthetic documents and their corresponding queries, we propose DocGen-RL, which regards the estimated relevance of the document as a reward and leverages reinforcement learning (RL) to optimize Doc-Gen pipeline.Extensive experiments demonstrate that DocGen and DocGen-RL significantly outperform existing state-of-the-art data augmentation methods, such as InPars, indicating that our new perspective of generating documents leverages the capacity of LLMs in generating synthetic data more effectively.We release the code, generated data, and model checkpoints to foster research in this area 1 .

Query performance prediction (QPP) is a core task in information retrieval. The QPP task is to predict the retrieval quality of a search system for a query without relevance judgments. Research has shown the effectiveness and usefulness of QPP for ad-hoc search. Recent years have witnessed considerable progress in conversational search (CS). Effective QPP could help a CS system to decide an appropriate action to be taken at the next turn. Despite its potential, QPP for CS has been little studied. We address this research gap by reproducing and studying the effectiveness of existing QPP methods in the context of CS. While the task of passage retrieval remains the same in the two settings, a user query in CS depends on the conversational history, introducing novel QPP challenges. In particular, we seek to explore to what extent findings from QPP methods for ad-hoc search generalize to three CS settings: (i) estimating the retrieval quality of different query rewriting-based retrieval methods, (ii) estimating the retrieval quality of a conversational dense retrieval method, and (iii) estimating the retrieval quality for top ranks vs. deeper-ranked lists. Our findings can be summarized as follows: (i) supervised QPP methods distinctly outperform unsupervised counterparts only when a large-scale training set is available; (ii) point-wise supervised QPP methods outperform their list-wise counterparts in most cases; and (iii) retrieval score-based unsupervised QPP methods show high effectiveness in assessing the conversational dense retrieval method, ConvDR.

In the knowledge-grounded conversation (KGC) task systems aim to produce more informative responses by leveraging external knowledge. KGC includes a vital part, knowledge selection, where conversational agents select the appropriate knowledge to be incorporated in the next response. Mixed initiative is an intrinsic feature of conversations where the user and the system can both take the initiative in suggesting new conversational directions. Knowledge selection can be driven by the user's initiative or by the system's initiative. For the former, the system usually selects knowledge according to the current user utterance that contains new topics or questions posed by the user; for the latter, the system usually selects knowledge according to the previously selected knowledge. No previous study has considered the mixed-initiative characteristics of knowledge selection to improve its performance.

Today's open-domain conversational agents increase the informativeness of generated responses by leveraging external knowledge. Most of the existing approaches work only for scenarios with a massive amount of monolingual knowledge sources. For languages with limited availability of knowledge sources, it is not effective to use knowledge in the same language to generate informative responses. To address this problem, we propose the task of cross-lingual knowledge grounded conversation (CKGC), where we leverage large-scale knowledge sources in another language to generate informative responses. Two main challenges come with the task of cross-lingual knowledge grounded conversation: (1) knowledge selection and response generation in a cross-lingual setting; and (2) the lack of a test dataset for evaluation. To tackle the first challenge, we propose the curriculum self-knowledge distillation (CSKD) scheme, which utilizes a large-scale dialogue corpus in an auxiliary language to improve cross-lingual knowledge selection and knowledge expression in the target language via knowledge distillation. To tackle the second challenge, we collect a cross-lingual knowledge grounded conversation test dataset to facilitate relevant research in the future. Extensive experiments on the newly created dataset verify the effectiveness of our proposed curriculum self-knowledge distillation method for cross-lingual knowledge grounded conversation. In addition, we find that our proposed unsupervised method significantly outperforms the state-of-the-art baselines in cross-lingual knowledge selection.

Existing conversational systems tend to generate generic responses. Recently, Background Based Conversation (BBCs) have been introduced to address this issue. Here, the generated responses are grounded in some background information. The proposed methods for BBCs are able to generate more informative responses, however, they either cannot generate natural responses or have difficulties in locating the right background information. In this paper, we propose a Reference-aware Network (RefNet) to address both issues. Unlike existing methods that generate responses token by token, RefNet incorporates a novel reference decoder that provides an alternative way to learn to directly select a semantic unit (e.g., a span containing complete semantic information) from the background. Experimental results show that RefNet significantly outperforms state-of-the-art methods in terms of both automatic and human evaluations, indicating that RefNet can generate more appropriate and human-like responses.

Today's conversational agents often generate responses that not sufficiently informative. One way of making them more informative is through the use of of external knowledge sources with so-called Knowledge-Grounded Conversations (KGCs). In this paper, we target the Knowledge Selection (KS) task, a key ingredient in KGC, that is aimed at selecting the appropriate knowledge to be used in the next response. Existing approaches to Knowledge Selection (KS) based on learned representations of the conversation context, that is previous conversation turns, and use Maximum Likelihood Estimation (MLE) to optimize KS. Such approaches have two main limitations. First, they do not explicitly track what knowledge has been used in the conversation nor how topics have shifted during the conversation. Second, MLE often relies on a limited set of example conversations for training, from which it is hard to infer that facts retrieved from the knowledge source can be re-used in multiple conversation contexts, and vice versa.