Claudia Hauff

Gustavo Penha, Claudia Hauff

Ranking responses for a given dialogue context is a popular benchmark in which the setup is to re-rank the ground-truth response over a limited set of $n$ responses, where $n$ is typically 10. The predominance of this setup in conversation response ranking has lead to a great deal of attention to building neural re-rankers, while the first-stage retrieval step has been overlooked. Since the correct answer is always available in the candidate list of $n$ responses, this artificial evaluation setup assumes that there is a first-stage retrieval step which is always able to rank the correct response in its top-$n$ list. In this paper we focus on the more realistic task of full-rank retrieval of responses, where $n$ can be up to millions of responses. We investigate both dialogue context and response expansion techniques for sparse retrieval, as well as zero-shot and fine-tuned dense retrieval approaches. Our findings based on three different information-seeking dialogue datasets reveal that a learned response expansion technique is a solid baseline for sparse retrieval. We find the best performing method overall to be dense retrieval with intermediate training, i.e. a step after the language model pre-training where sentence representations are learned, followed by fine-tuning on the target conversational data. We also investigate the intriguing phenomena that harder negatives sampling techniques lead to worse results for the fine-tuned dense retrieval models. The code and datasets are available at https://github.com/Guzpenha/transformer_rankers/tree/full_rank_retrieval_dialogues.

Gustavo Penha, Arthur Câmara, Claudia Hauff

Heavily pre-trained transformers for language modelling, such as BERT, have shown to be remarkably effective for Information Retrieval (IR) tasks, typically applied to re-rank the results of a first-stage retrieval model. IR benchmarks evaluate the effectiveness of retrieval pipelines based on the premise that a single query is used to instantiate the underlying information need. However, previous research has shown that (I) queries generated by users for a fixed information need are extremely variable and, in particular, (II) neural models are brittle and often make mistakes when tested with modified inputs. Motivated by those observations we aim to answer the following question: how robust are retrieval pipelines with respect to different variations in queries that do not change the queries' semantics? In order to obtain queries that are representative of users' querying variability, we first created a taxonomy based on the manual annotation of transformations occurring in a dataset (UQV100) of user-created query variations. For each syntax-changing category of our taxonomy, we employed different automatic methods that when applied to a query generate a query variation. Our experimental results across two datasets for two IR tasks reveal that retrieval pipelines are not robust to these query variations, with effectiveness drops of $\approx20\%$ on average. The code and datasets are available at https://github.com/Guzpenha/query_variation_generators.

Gustavo Penha, Claudia Hauff

We study Label Smoothing (LS), a widely used regularization technique, in the context of neural learning to rank (L2R) models. LS combines the ground-truth labels with a uniform distribution, encouraging the model to be less confident in its predictions. We analyze the relationship between the non-relevant documents-specifically how they are sampled-and the effectiveness of LS, discussing how LS can be capturing "hidden similarity knowledge" between the relevantand non-relevant document classes. We further analyze LS by testing if a curriculum-learning approach, i.e., starting with LS and after anumber of iterations using only ground-truth labels, is beneficial. Inspired by our investigation of LS in the context of neural L2R models, we propose a novel technique called Weakly Supervised Label Smoothing (WSLS) that takes advantage of the retrieval scores of the negative sampled documents as a weak supervision signal in the process of modifying the ground-truth labels. WSLS is simple to implement, requiring no modification to the neural ranker architecture. Our experiments across three retrieval tasks-passage retrieval, similar question retrieval and conversation response ranking-show that WSLS for pointwise BERT-based rankers leads to consistent effectiveness gains. The source code is available at https://anonymous.4open.science/r/dac85d48-6f71-4261-a7d8-040da6021c52/.

Gustavo Penha, Claudia Hauff

Understanding when and why neural ranking models fail for an IR task via error analysis is an important part of the research cycle. Here we focus on the challenges of (i) identifying categories of difficult instances (a pair of question and response candidates) for which a neural ranker is ineffective and (ii) improving neural ranking for such instances. To address both challenges we resort to slice-based learning for which the goal is to improve effectiveness of neural models for slices (subsets) of data. We address challenge (i) by proposing different slicing functions (SFs) that select slices of the dataset---based on prior work we heuristically capture different failures of neural rankers. Then, for challenge (ii) we adapt a neural ranking model to learn slice-aware representations, i.e. the adapted model learns to represent the question and responses differently based on the model's prediction of which slices they belong to. Our experimental results (the source code and data are available at https://github.com/Guzpenha/slice_based_learning) across three different ranking tasks and four corpora show that slice-based learning improves the effectiveness by an average of 2% over a neural ranker that is not slice-aware.

Maria Movin, Claudia Hauff, Aron Henriksson et al.

LLM-driven GUI agents are increasingly used in production systems to automate workflows and simulate users for evaluation and optimization. Yet most GUI-agent evaluations emphasize task success and provide limited evidence on whether agents interact in human-like ways. We present a trace-level evaluation framework that compares human and agent behavior across (i) task outcome and effort, (ii) query formulation, and (iii) navigation across interface states. We instantiate the framework in a controlled study in a production audio-streaming search application, where 39 participants and a state-of-the-art GUI agent perform ten multi-hop search tasks. The agent achieves task success comparable to participants and generates broadly aligned queries, but follows systematically different navigation strategies: participants exhibit content-centric, exploratory behavior, while the agent is more search-centric and low-branching. These results show that outcome and query alignment do not imply behavioral alignment, motivating trace-level diagnostics when deploying GUI agents as proxies for users in production search systems.

Konstantina Palla, José Luis Redondo García, Claudia Hauff et al.

Content moderation plays a critical role in shaping safe and inclusive online environments, balancing platform standards, user expectations, and regulatory frameworks. Traditionally, this process involves operationalising policies into guidelines, which are then used by downstream human moderators for enforcement, or to further annotate datasets for training machine learning moderation models. However, recent advancements in large language models (LLMs) are transforming this landscape. These models can now interpret policies directly as textual inputs, eliminating the need for extensive data curation. This approach offers unprecedented flexibility, as moderation can be dynamically adjusted through natural language interactions. This paradigm shift raises important questions about how policies are operationalised and the implications for content moderation practices. In this paper, we formalise the emerging policy-as-prompt framework and identify five key challenges across four domains: Technical Implementation (1. translating policy to prompts, 2. sensitivity to prompt structure and formatting), Sociotechnical (3. the risk of technological determinism in policy formation), Organisational (4. evolving roles between policy and machine learning teams), and Governance (5. model governance and accountability). Through analysing these challenges across technical, sociotechnical, organisational, and governance dimensions, we discuss potential mitigation approaches. This research provides actionable insights for practitioners and lays the groundwork for future exploration of scalable and adaptive content moderation systems in digital ecosystems.

Azin Ghazimatin, Ekaterina Garmash, Gustavo Penha et al.

Listeners of long-form talk-audio content, such as podcast episodes, often find it challenging to understand the overall structure and locate relevant sections. A practical solution is to divide episodes into chapters--semantically coherent segments labeled with titles and timestamps. Since most episodes on our platform at Spotify currently lack creator-provided chapters, automating the creation of chapters is essential. Scaling the chapterization of podcast episodes presents unique challenges. First, episodes tend to be less structured than written texts, featuring spontaneous discussions with nuanced transitions. Second, the transcripts are usually lengthy, averaging about 16,000 tokens, which necessitates efficient processing that can preserve context. To address these challenges, we introduce PODTILE, a fine-tuned encoder-decoder transformer to segment conversational data. The model simultaneously generates chapter transitions and titles for the input transcript. To preserve context, each input text is augmented with global context, including the episode's title, description, and previous chapter titles. In our intrinsic evaluation, PODTILE achieved an 11% improvement in ROUGE score over the strongest baseline. Additionally, we provide insights into the practical benefits of auto-generated chapters for listeners navigating episode content. Our findings indicate that auto-generated chapters serve as a useful tool for engaging with less popular podcasts. Finally, we present empirical evidence that using chapter titles can enhance effectiveness of sparse retrieval in search tasks.

Ali Vardasbi, Gustavo Penha, Claudia Hauff et al.

When using LLMs to rank items based on given criteria, or evaluate answers, the order of candidate items can influence the model's final decision. This sensitivity to item positioning in a LLM's prompt is known as position bias. Prior research shows that this bias exists even in large models, though its severity varies across models and tasks. In addition to position bias, LLMs also exhibit varying degrees of low repetition consistency, where repeating the LLM call with the same candidate ordering can lead to different rankings. To address both inconsistencies, a common approach is to prompt the model multiple times with different candidate orderings and aggregate the results via majority voting. However, this repetition strategy, significantly increases computational costs. Extending prior findings, we observe that both the direction -- favoring either the earlier or later candidate in the prompt -- and magnitude of position bias across instances vary substantially, even within a single dataset. This observation highlights the need for a per-instance mitigation strategy. To this end, we introduce a dynamic early-stopping method that adaptively determines the number of repetitions required for each instance. Evaluating our approach across three LLMs of varying sizes and on two tasks, namely re-ranking and alignment, we demonstrate that transitioning to a dynamic repetition strategy reduces the number of LLM calls by an average of 81%, while preserving the accuracy. Furthermore, we propose a confidence-based adaptation to our early-stopping method, reducing LLM calls by an average of 87% compared to static repetition, with only a slight accuracy trade-off relative to our original early-stopping method.

Arthur Câmara, David Maxwell, Claudia Hauff

Complex search tasks - such as those from the Search as Learning (SAL) domain - often result in users developing an information need composed of several aspects. However, current models of searcher behaviour assume that individuals have an atomic need, regardless of the task. While these models generally work well for simpler informational needs, we argue that searcher models need to be developed further to allow for the decomposition of a complex search task into multiple aspects. As no searcher model yet exists that considers both aspects and the SAL domain, we propose, by augmenting the Complex Searcher Model (CSM), the Subtopic Aware Complex Searcher Model (SACSM) - modelling aspects as subtopics to the user's need. We then instantiate several agents (i.e., simulated users), with different subtopic selection strategies, which can be considered as different prototypical learning strategies (e.g., should I deeply examine one subtopic at a time, or shallowly cover several subtopics?). Finally, we report on the first large-scale simulated analysis of user behaviours in the SAL domain. Results demonstrate that the SACSM, under certain conditions, simulates user behaviours accurately.

Arthur Câmara, Claudia Hauff

Word embeddings, made widely popular in 2013 with the release of word2vec, have become a mainstay of NLP engineering pipelines. Recently, with the release of BERT, word embeddings have moved from the term-based embedding space to the contextual embedding space -- each term is no longer represented by a single low-dimensional vector but instead each term and \emph{its context} determine the vector weights. BERT's setup and architecture have been shown to be general enough to be applicable to many natural language tasks. Importantly for Information Retrieval (IR), in contrast to prior deep learning solutions to IR problems which required significant tuning of neural net architectures and training regimes, "vanilla BERT" has been shown to outperform existing retrieval algorithms by a wide margin, including on tasks and corpora that have long resisted retrieval effectiveness gains over traditional IR baselines (such as Robust04). In this paper, we employ the recently proposed axiomatic dataset analysis technique -- that is, we create diagnostic datasets that each fulfil a retrieval heuristic (both term matching and semantic-based) -- to explore what BERT is able to learn. In contrast to our expectations, we find BERT, when applied to a recently released large-scale web corpus with ad-hoc topics, to \emph{not} adhere to any of the explored axioms. At the same time, BERT outperforms the traditional query likelihood retrieval model by 40\%. This means that the axiomatic approach to IR (and its extension of diagnostic datasets created for retrieval heuristics) may in its current form not be applicable to large-scale corpora. Additional -- different -- axioms are needed.

Arthur Câmara, Nirmal Roy, David Maxwell et al.

Search engines are considered the primary tool to assist and empower learners in finding information relevant to their learning goals-be it learning something new, improving their existing skills, or just fulfilling a curiosity. While several approaches for improving search engines for the learning scenario have been proposed, instructional scaffolding has not been studied in the context of search as learning, despite being shown to be effective for improving learning in both digital and traditional learning contexts. When scaffolding is employed, instructors provide learners with support throughout their autonomous learning process. We hypothesize that the usage of scaffolding techniques within a search system can be an effective way to help learners achieve their learning objectives whilst searching. As such, this paper investigates the incorporation of scaffolding into a search system employing three different strategies (as well as a control condition): (I) AQE_{SC}, the automatic expansion of user queries with relevant subtopics; (ii) CURATED_{SC}, the presenting of a manually curated static list of relevant subtopics on the search engine result page; and (iii) FEEDBACK_{SC}, which projects real-time feedback about a user's exploration of the topic space on top of the CURATED_{SC} visualization. To investigate the effectiveness of these approaches with respect to human learning, we conduct a user study (N=126) where participants were tasked with searching and learning about topics such as `genetically modified organisms'. We find that (I) the introduction of the proposed scaffolding methods does not significantly improve learning gains. However, (ii) it does significantly impact search behavior. Furthermore, (iii) immediate feedback of the participants' learning leads to undesirable user behavior, with participants focusing on the feedback gauges instead of learning.

Gustavo Penha, Claudia Hauff

According to the Probability Ranking Principle (PRP), ranking documents in decreasing order of their probability of relevance leads to an optimal document ranking for ad-hoc retrieval. The PRP holds when two conditions are met: [C1] the models are well calibrated, and, [C2] the probabilities of relevance are reported with certainty. We know however that deep neural networks (DNNs) are often not well calibrated and have several sources of uncertainty, and thus [C1] and [C2] might not be satisfied by neural rankers. Given the success of neural Learning to Rank (L2R) approaches-and here, especially BERT-based approaches-we first analyze under which circumstances deterministic, i.e. outputs point estimates, neural rankers are calibrated. Then, motivated by our findings we use two techniques to model the uncertainty of neural rankers leading to the proposed stochastic rankers, which output a predictive distribution of relevance as opposed to point estimates. Our experimental results on the ad-hoc retrieval task of conversation response ranking reveal that (i) BERT-based rankers are not robustly calibrated and that stochastic BERT-based rankers yield better calibration; and (ii) uncertainty estimation is beneficial for both risk-aware neural ranking, i.e.taking into account the uncertainty when ranking documents, and for predicting unanswerable conversational contexts.

Gustavo Penha, Claudia Hauff

Heavily pre-trained transformer models such as BERT have recently shown to be remarkably powerful at language modelling by achieving impressive results on numerous downstream tasks. It has also been shown that they are able to implicitly store factual knowledge in their parameters after pre-training. Understanding what the pre-training procedure of LMs actually learns is a crucial step for using and improving them for Conversational Recommender Systems (CRS). We first study how much off-the-shelf pre-trained BERT "knows" about recommendation items such as books, movies and music. In order to analyze the knowledge stored in BERT's parameters, we use different probes that require different types of knowledge to solve, namely content-based and collaborative-based. Content-based knowledge is knowledge that requires the model to match the titles of items with their content information, such as textual descriptions and genres. In contrast, collaborative-based knowledge requires the model to match items with similar ones, according to community interactions such as ratings. We resort to BERT's Masked Language Modelling head to probe its knowledge about the genre of items, with cloze style prompts. In addition, we employ BERT's Next Sentence Prediction head and representations' similarity to compare relevant and non-relevant search and recommendation query-document inputs to explore whether BERT can, without any fine-tuning, rank relevant items first. Finally, we study how BERT performs in a conversational recommendation downstream task. Overall, our analyses and experiments show that: (i) BERT has knowledge stored in its parameters about the content of books, movies and music; (ii) it has more content-based knowledge than collaborative-based knowledge; and (iii) fails on conversational recommendation when faced with adversarial data.

Gustavo Penha, Claudia Hauff

Neural ranking models are traditionally trained on a series of random batches, sampled uniformly from the entire training set. Curriculum learning has recently been shown to improve neural models' effectiveness by sampling batches non-uniformly, going from easy to difficult instances during training. In the context of neural Information Retrieval (IR) curriculum learning has not been explored yet, and so it remains unclear (1) how to measure the difficulty of training instances and (2) how to transition from easy to difficult instances during training. To address both challenges and determine whether curriculum learning is beneficial for neural ranking models, we need large-scale datasets and a retrieval task that allows us to conduct a wide range of experiments. For this purpose, we resort to the task of conversation response ranking: ranking responses given the conversation history. In order to deal with challenge (1), we explore scoring functions to measure the difficulty of conversations based on different input spaces. To address challenge (2) we evaluate different pacing functions, which determine the velocity in which we go from easy to difficult instances. We find that, overall, by just intelligently sorting the training data (i.e., by performing curriculum learning) we can improve the retrieval effectiveness by up to 2%.

Gustavo Penha, Alexandru Balan, Claudia Hauff

Conversational search is an approach to information retrieval (IR), where users engage in a dialogue with an agent in order to satisfy their information needs. Previous conceptual work described properties and actions a good agent should exhibit. Unlike them, we present a novel conceptual model defined in terms of conversational goals, which enables us to reason about current research practices in conversational search. Based on the literature, we elicit how existing tasks and test collections from the fields of IR, natural language processing (NLP) and dialogue systems (DS) fit into this model. We describe a set of characteristics that an ideal conversational search dataset should have. Lastly, we introduce MANtIS (the code and dataset are available at https://guzpenha.github.io/MANtIS/), a large-scale dataset containing multi-domain and grounded information seeking dialogues that fulfill all of our dataset desiderata. We provide baseline results for the conversation response ranking and user intent prediction tasks.