Test collections are crucial for evaluating Information Retrieval (IR) systems. Creating a diverse set of user queries for these collections can be challenging, and obtaining relevance judgments, which indicate how well retrieved documents match a query, is often costly and resource-intensive. Recently, generating synthetic datasets using Large Language Models (LLMs) has gained attention in various applications. While previous work has used LLMs to generate synthetic queries or documents to improve ranking models, using LLMs to create synthetic test collections is still relatively unexplored. Previous work~\cite{rahmani2024synthetic} showed that synthetic test collections have the potential to be used for system evaluation, however, more analysis is needed to validate this claim. In this paper, we thoroughly investigate the reliability of synthetic test collections constructed using LLMs, where LLMs are used to generate synthetic queries, labels, or both. In particular, we examine the potential biases that might occur when such test collections are used for evaluation. We first empirically show the presence of such bias in evaluation results and analyse the effects it might have on system evaluation. We further validate the presence of such bias using a linear mixed-effects model. Our analysis shows that while the effect of bias present in evaluation results obtained using synthetic test collections could be significant, for e.g.~computing absolute system performance, its effect may not be as significant in comparing relative system performance. Codes and data are available at: https://github.com/rahmanidashti/BiasSyntheticData.

Neural IR architectures, particularly cross-encoders, are highly effective models whose internal mechanisms are mostly unknown. Most works trying to explain their behavior focused on high-level processes (e.g., what in the input influences the prediction, does the model adhere to known IR axioms) but fall short of describing the matching process. Instead of Mechanistic Interpretability approaches which specifically aim at explaining the hidden mechanisms of neural models, we demonstrate that more straightforward methods can already provide valuable insights. In this paper, we first focus on the attention process and extract causal insights highlighting the crucial roles of some attention heads in this process. Second, we provide an interpretation of the mechanism underlying matching detection.

This is the fifth year of the TREC Deep Learning track. As in previous years, we leverage the MS MARCO datasets that made hundreds of thousands of human-annotated training labels available for both passage and document ranking tasks. We mostly repeated last year's design, to get another matching test set, based on the larger, cleaner, less-biased v2 passage and document set, with passage ranking as primary and document ranking as a secondary task (using labels inferred from passage). As we did last year, we sample from MS MARCO queries that were completely held out, unused in corpus construction, unlike the test queries in the first three years. This approach yields a more difficult test with more headroom for improvement. Alongside the usual MS MARCO (human) queries from MS MARCO, this year we generated synthetic queries using a fine-tuned T5 model and using a GPT-4 prompt. The new headline result this year is that runs using Large Language Model (LLM) prompting in some way outperformed runs that use the "nnlm" approach, which was the best approach in the previous four years. Since this is the last year of the track, future iterations of prompt-based ranking can happen in other tracks. Human relevance assessments were applied to all query types, not just human MS MARCO queries. Evaluation using synthetic queries gave similar results to human queries, with system ordering agreement of $τ=0.8487$. However, human effort was needed to select a subset of the synthetic queries that were usable. We did not see clear evidence of bias, where runs using GPT-4 were favored when evaluated using synthetic GPT-4 queries, or where runs using T5 were favored when evaluated on synthetic T5 queries.

Large Language Models (LLMs) are increasingly used to automate relevance judgments for information retrieval (IR) tasks, often demonstrating agreement with human labels that approaches inter-human agreement. To assess the robustness and reliability of LLM-based relevance judgments, we systematically investigate impact of prompt sensitivity on the task. We collected prompts for relevance assessment from 15 human experts and 15 LLMs across three tasks~ -- ~binary, graded, and pairwise~ -- ~yielding 90 prompts in total. After filtering out unusable prompts from three humans and three LLMs, we employed the remaining 72 prompts with three different LLMs as judges to label document/query pairs from two TREC Deep Learning Datasets (2020 and 2021). We compare LLM-generated labels with TREC official human labels using Cohen's $κ$ and pairwise agreement measures. In addition to investigating the impact of prompt variations on agreement with human labels, we compare human- and LLM-generated prompts and analyze differences among different LLMs as judges. We also compare human- and LLM-generated prompts with the standard UMBRELA prompt used for relevance assessment by Bing and TREC 2024 Retrieval Augmented Generation (RAG) Track. To support future research in LLM-based evaluation, we release all data and prompts at https://github.com/Narabzad/prompt-sensitivity-relevance-judgements/.

Fine-tuning in information retrieval systems using pre-trained language models (PLM-based IR) requires learning query representations and query-document relations, in addition to downstream task-specific learning. This study introduces coarse-tuning as an intermediate learning stage that bridges pre-training and fine-tuning. By learning query representations and query-document relations in coarse-tuning, we aim to reduce the load of fine-tuning and improve the learning effect of downstream IR tasks. We propose Query-Document Pair Prediction (QDPP) for coarse-tuning, which predicts the appropriateness of query-document pairs. Evaluation experiments show that the proposed method significantly improves MRR and/or nDCG@5 in four ad-hoc document retrieval datasets. Furthermore, the results of the query prediction task suggested that coarse-tuning facilitated learning of query representation and query-document relations.

Relevance labels, which indicate whether a search result is valuable to a searcher, are key to evaluating and optimising search systems. The best way to capture the true preferences of users is to ask them for their careful feedback on which results would be useful, but this approach does not scale to produce a large number of labels. Getting relevance labels at scale is usually done with third-party labellers, who judge on behalf of the user, but there is a risk of low-quality data if the labeller doesn't understand user needs. To improve quality, one standard approach is to study real users through interviews, user studies and direct feedback, find areas where labels are systematically disagreeing with users, then educate labellers about user needs through judging guidelines, training and monitoring. This paper introduces an alternate approach for improving label quality. It takes careful feedback from real users, which by definition is the highest-quality first-party gold data that can be derived, and develops an large language model prompt that agrees with that data. We present ideas and observations from deploying language models for large-scale relevance labelling at Bing, and illustrate with data from TREC. We have found large language models can be effective, with accuracy as good as human labellers and similar capability to pick the hardest queries, best runs, and best groups. Systematic changes to the prompts make a difference in accuracy, but so too do simple paraphrases. To measure agreement with real searchers needs high-quality ``gold'' labels, but with these we find that models produce better labels than third-party workers, for a fraction of the cost, and these labels let us train notably better rankers.

Evaluation efforts such as TREC, CLEF, NTCIR and FIRE, alongside public leaderboard such as MS MARCO, are intended to encourage research and track our progress, addressing big questions in our field. However, the goal is not simply to identify which run is "best", achieving the top score. The goal is to move the field forward by developing new robust techniques, that work in many different settings, and are adopted in research and practice. This paper uses the MS MARCO and TREC Deep Learning Track as our case study, comparing it to the case of TREC ad hoc ranking in the 1990s. We show how the design of the evaluation effort can encourage or discourage certain outcomes, and raising questions about internal and external validity of results. We provide some analysis of certain pitfalls, and a statement of best practices for avoiding such pitfalls. We summarize the progress of the effort so far, and describe our desired end state of "robust usefulness", along with steps that might be required to get us there.

The TREC Deep Learning (DL) Track studies ad hoc search in the large data regime, meaning that a large set of human-labeled training data is available. Results so far indicate that the best models with large data may be deep neural networks. This paper supports the reuse of the TREC DL test collections in three ways. First we describe the data sets in detail, documenting clearly and in one place some details that are otherwise scattered in track guidelines, overview papers and in our associated MS MARCO leaderboard pages. We intend this description to make it easy for newcomers to use the TREC DL data. Second, because there is some risk of iteration and selection bias when reusing a data set, we describe the best practices for writing a paper using TREC DL data, without overfitting. We provide some illustrative analysis. Finally we address a number of issues around the TREC DL data, including an analysis of reusability.

The Transformer-Kernel (TK) model has demonstrated strong reranking performance on the TREC Deep Learning benchmark -- and can be considered to be an efficient (but slightly less effective) alternative to other Transformer-based architectures that employ (i) large-scale pretraining (high training cost), (ii) joint encoding of query and document (high inference cost), and (iii) larger number of Transformer layers (both high training and high inference costs). Since, a variant of the TK model -- called TKL -- has been developed that incorporates local self-attention to efficiently process longer input sequences in the context of document ranking. In this work, we propose a novel Conformer layer as an alternative approach to scale TK to longer input sequences. Furthermore, we incorporate query term independence and explicit term matching to extend the model to the full retrieval setting. We benchmark our models under the strictly blind evaluation setting of the TREC 2020 Deep Learning track and find that our proposed architecture changes lead to improved retrieval quality over TKL. Our best model also outperforms all non-neural runs ("trad") and two-thirds of the pretrained Transformer-based runs ("nnlm") on NDCG@10.

The Conformer-Kernel (CK) model [Mitra et al., 2020] builds upon the Transformer-Kernel (TK) [Hofstätter et al., 2019] architecture, that demonstrated strong competitive performance compared to BERTbased [Devlin et al., 2019] ranking methods, but notably at a fraction of the compute and GPU memory cost, at the TREC 2019 Deep Learning track [Craswell et al., 2020b]. Notwithstanding these strong results, the TK model suffers from two clear deficiencies. Firstly, because the TK model employs stacked Transformers for query and document encoding, it is challenging to incorporate long body text into this model as the GPU memory requirement of Transformers' self-attention layers grows quadratically with respect to input sequence length. So, for example, to increase the limit on the maximum input sequence length by 4 from 128 to 512 we would require 16 more GPU memory for each of the self-attention layers in the model. Considering that documents can contain thousands of terms, this limits the model to inspecting only a subset of the document text which may have negative implications, such as poorer retrieval quality and under-retrieval of longer documents [Hofstätter et al., 2020].