Context-grounded hallucinations are cases where model outputs contain information not verifiable against the source text. We study the applicability of LLMs for localizing such hallucinations, as a more practical alternative to existing complex evaluation pipelines. In the absence of established benchmarks for meta-evaluation of hallucinations localization, we construct one tailored to LLMs, involving a challenging human annotation of over 1,000 examples. We complement the benchmark with an LLM-based evaluation protocol, verifying its quality in a human evaluation. Since existing representations of hallucinations limit the types of errors that can be expressed, we propose a new representation based on free-form textual descriptions, capturing the full range of possible errors. We conduct a comprehensive study, evaluating four large-scale LLMs, which highlights the benchmark’s difficulty, as the best model achieves an F1 score of only 0.67. Through careful analysis, we offer insights into optimal prompting strategies for the task and identify the main factors that make it challenging for LLMs: (1) a tendency to incorrectly flag missing details as inconsistent, despite being instructed to check only facts in the output; and (2) difficulty with outputs containing factually correct information absent from the source - and thus not verifiable - due to alignment with the model’s parametric knowledge.

Model merging has emerged as a promising technique for combining multiple fine-tuned models into a single multitask model without retraining. However, the factors that determine whether merging will succeed or fail remain poorly understood. In this work, we investigate why specific models are merged better than others. To do so, we propose a concrete, measurable definition of mergeability. We investigate several potential causes for high or low mergeability, highlighting the base model knowledge as a dominant factor: Models fine-tuned on instances that the base model knows better are more mergeable than models fine-tuned on instances that the base model struggles with. Based on our mergeability definition, we explore a simple weighted merging technique that better preserves weak knowledge in the base model.

We deal with the problem of Question Answering (QA) over a long document, which poses a challenge for modern Large Language Models (LLMs). Although LLMs can handle increasingly longer context windows, they struggle to effectively utilize the long content. To address this issue, we introduce the concept of a virtual document (VDoc). A VDoc is created by selecting chunks from the original document that are most likely to contain the information needed to answer the user’s question, while ensuring they fit within the LLM’s context window. We hypothesize that providing a short and focused VDoc to the LLM is more effective than filling the entire context window with less relevant information. Our experiments confirm this hypothesis and demonstrate that using VDocs improves results on the QA task.

We deal with the scenario of conversational search, where user queries are under-specified or ambiguous. This calls for a mixed-initiative setup. User-asks (queries) and system-answers, as well as system-asks (clarification questions) and user response, in order to clarify her information needs. We focus on the task of selecting the next clarification question, given conversation context. Our method leverages passage retrieval from background content to fine-tune two deep-learning models for ranking candidate clarification questions. We evaluated our method on two different use-cases. The first is an open domain conversational search in a large web collection. The second is a task-oriented customer-support setup. We show that our method performs well on both use-cases.

Customer support agents play a crucial role as an interface between an organization and its end-users. We propose CAIRAA: Conversational Approach to Information Retrieval for Agent Assistance, to reduce the cognitive workload of support agents who engage with users through conversation systems. CAIRAA monitors an evolving conversation and recommends both responses and URLs of documents the agent can use in replies to their client. We combine traditional information retrieval (IR) approaches with more recent Deep Learning (DL) models to ensure high accuracy and efficient run-time performance in the deployed system. Here, we describe the CAIRAA system and demonstrate its effectiveness in a pilot study via a short video.

A frequent pattern in customer care conversations is the agents responding with appropriate webpage URLs that address users’ needs. We study the task of predicting the documents that customer care agents can use to facilitate users’ needs. We also introduce a new public dataset which supports the aforementioned problem. Using this dataset and two others, we investigate state-of-the art deep learning (DL) and information retrieval (IR) models for the task. Additionally, we analyze the practicality of such systems in terms of inference time complexity. Our show that an hybrid IR+DL approach provides the best of both worlds.

We focus on the task of Frequently Asked Questions (FAQ) retrieval. A given user query can be matched against the questions and/or the answers in the FAQ. We present a fully unsupervised method that exploits the FAQ pairs to train two BERT models. The two models match user queries to FAQ answers and questions, respectively. We alleviate the missing labeled data of the latter by automatically generating high-quality question paraphrases. We show that our model is on par and even outperforms supervised models on existing datasets.

We describe a weakly-supervised method for training deep learning models for the task of ad-hoc document retrieval. Our method is based on generative and discriminative models that are trained using weak-supervision just from the documents in the corpus. We present an end-to-end retrieval system that starts with traditional information retrieval methods, followed by two deep learning re-rankers. We evaluate our method on three different datasets: a COVID-19 related scientific literature dataset and two news datasets. We show that our method outperforms state-of-the-art methods; this without the need for the expensive process of manually labeling data.

We present a novel system providing summaries for Computer Science publications. Through a qualitative user study, we identified the most valuable scenarios for discovery, exploration and understanding of scientific documents. Based on these findings, we built a system that retrieves and summarizes scientific documents for a given information need, either in form of a free-text query or by choosing categorized values such as scientific tasks, datasets and more. Our system ingested 270,000 papers, and its summarization module aims to generate concise yet detailed summaries. We validated our approach with human experts.

In this paper we suggest to leverage the partition of articles into sections, in order to learn thematic similarity metric between sentences. We assume that a sentence is thematically closer to sentences within its section than to sentences from other sections. Based on this assumption, we use Wikipedia articles to automatically create a large dataset of weakly labeled sentence triplets, composed of a pivot sentence, one sentence from the same section and one from another section. We train a triplet network to embed sentences from the same section closer. To test the performance of the learned embeddings, we create and release a sentence clustering benchmark. We show that the triplet network learns useful thematic metrics, that significantly outperform state-of-the-art semantic similarity methods and multipurpose embeddings on the task of thematic clustering of sentences. We also show that the learned embeddings perform well on the task of sentence semantic similarity prediction.