Neural IR systems often employ a retrieve-and-rerank framework: a bi-encoder retrieves a fixed number of candidates (e.g., 𝐾=100), which a cross-encoder then reranks.Recent studies have indicated that relevance feedback from the reranker at inference time can improve the recall of the retriever.The approach works by updating the retriever's query representations via a distillation process that aligns it with the reranker's predictions.While a powerful idea, the arguably narrow scope of past studies focusing on a small number of specific domains such as english question answering and entity retrieval has left a gap in our understanding of how well it generalizes.In this paper, we study inference-time reranker relevance feedback extensively across multiple retrieval domains, languages, and modalities, while also investigating aspects such as the performance and latency implications of the number of distillation updates and feedback candidates.

Managing the data for Information Retrieval (IR) experiments can be challenging. Dataset documentation is scattered across the Internet and once one obtains a copy of the data, there are numerous different data formats to work with. Even basic formats can have subtle dataset-specific nuances that need to be considered for proper use. To help mitigate these challenges, we introduce a new robust and lightweight tool (ir_datasets) for acquiring, managing, and performing typical operations over datasets used in IR. We primarily focus on textual datasets used for ad-hoc search. This tool provides both a Python and command line interface to numerous IR datasets and benchmarks. To our knowledge, this is the most extensive tool of its kind. Integrations with popular IR indexing and experimentation toolkits demonstrate the tool's utility. We also provide documentation of these datasets through the \sys catalog: https://ir-datasets.com/. The catalog acts as a hub for information on datasets used in IR, providing core information about what data each benchmark provides as well as links to more detailed information. We welcome community contributions and intend to continue to maintain and grow this tool.

Automatically generating accurate summaries from clinical reports could save a clinician's time, improve summary coverage, and reduce errors. We propose a sequence-to-sequence abstractive summarization model augmented with domain-specific ontological information to enhance content selection and summary generation. We apply our method to a dataset of radiology reports and show that it significantly outperforms the current state-of-the-art on this task in terms of rouge scores. Extensive human evaluation conducted by a radiologist further indicates that this approach yields summaries that are less likely to omit important details, without sacrificing readability or accuracy.

Although considerable attention has been given to neural ranking architectures recently, far less attention has been paid to the term representations that are used as input to these models. In this work, we investigate how two pretrained contextualized language models (ELMo and BERT) can be utilized for ad-hoc document ranking. Through experiments on TREC benchmarks, we find that several ex-sting neural ranking architectures can benefit from the additional context provided by contextualized language models. Furthermore, we propose a joint approach that incorporates BERT's classification vector into existing neural models and show that it outperforms state-of-the-art ad-hoc ranking baselines. We call this joint approach CEDR (Contextualized Embeddings for Document Ranking). We also address practical challenges in using these models for ranking, including the maximum input length imposed by BERT and runtime performance impacts of contextualized language models.

Complex answer retrieval (CAR) is the process of retrieving answers to questions that have multifaceted or nuanced answers. In this work, we present two novel approaches for CAR based on the observation that question facets can vary in utility: from structural (facets that can apply to many similar topics, such as 'History') to topical (facets that are specific to the question's topic, such as the 'Westward expansion' of the United States). We first explore a way to incorporate facet utility into ranking models during query term score combination. We then explore a general approach to reform the structure of ranking models to aid in learning of facet utility in the query-document term matching phase. When we use our techniques with a leading neural ranker on the TREC CAR dataset, our methods yield statistically significant improvements over both an unmodified neural architecture and submitted TREC runs.

Citation texts are sometimes not very informative or in some cases inaccurate by themselves; they need the appropriate context from the referenced paper to reflect its exact contributions. To address this problem, we propose an unsupervised model that uses distributed representation of words as well as domain knowledge to extract the appropriate context from the reference paper. Evaluation results show the effectiveness of our model by significantly outperforming the state-of-the-art. We furthermore demonstrate how an effective contextualization method results in improving citation-based summarization of the scientific articles.

In recent years, social media has become a significant resource for improving healthcare and mental health. Mental health forums are online communities where people express their issues, and seek help from moderators and other users. In such forums, there are often posts with severe content indicating that the user is in acute distress and there is a risk of attempted self‐harm. Moderators need to respond to these severe posts in a timely manner to prevent potential self‐harm. However, the large volume of daily posted content makes it difficult for the moderators to locate and respond to these critical posts. We propose an approach for triaging user content into four severity categories that are defined based on an indication of self‐harm ideation. Our models are based on a feature‐rich classification framework, which includes lexical, psycholinguistic, contextual, and topic modeling features. Our approaches improve over the state of the art in triaging the content severity in mental health forums by large margins (up to 17% improvement over the F‐1 scores). Furthermore, using our proposed model, we analyze the mental state of users and we show that overall, long‐term users of the forum demonstrate decreased severity of risk over time. Our analysis on the interaction of the moderators with the users further indicates that without an automatic way to identify critical content, it is indeed challenging for the moderators to provide timely response to the users in need.