Papers by Yosi Mass
Fine-Grained Detection of Context-Grounded Hallucinations Using LLMs (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing representations of hallucinations limit the types of errors that can be expressed, so we propose a new representation based on free-form textual descriptions, capturing the full range of possible errors. |
| Approach: | They propose a benchmark for localizing hallucinations using LLMs with a human annotation of over 1,000 examples and a protocol to verify its quality in a humans evaluation. |
| Outcome: | The proposed representation captures the full range of possible errors, and the best model achieves an F1 score of 0.67. |
Learning Thematic Similarity Metric from Article Sections Using Triplet Networks (P18-2)
Copied to clipboard
| Challenge: | In this paper, we use Wikipedia articles to learn thematic similarity metric between sentences. |
| Approach: | They propose to leverage the partition of articles into sections to learn thematic similarity metric between sentences. |
| Outcome: | The proposed model outperforms state-of-the-art embeddings on the task of thematic clustering of sentences. |
Agent Assist through Conversation Analysis (2020.emnlp-demos)
Copied to clipboard
Kshitij Fadnis, Nathaniel Mills, Jatin Ganhotra, Haggai Roitman, Gaurav Pandey, Doron Cohen, Yosi Mass, Shai Erera, Chulaka Gunasekara, Danish Contractor, Siva Patel, Q. Vera Liao, Sachindra Joshi, Luis Lastras, David Konopnicki
| Challenge: | Using conversational approach to information retrieval for agent assistance, customer support agents are a critical part of an organization's customer support team. |
| Approach: | They propose a conversational approach to information retrieval for agent assistance that monitors an evolving conversation and recommends both responses and URLs of documents. |
| Outcome: | The proposed system monitors an evolving conversation and recommends both responses and URLs of documents the agent can use in replies to their client. |
Semantic Relatedness of Wikipedia Concepts – Benchmark Data and a Working Solution (L18-1)
Copied to clipboard
| Challenge: | Existing methods to measure relatedness between Wikipedia concepts are lacking. |
| Approach: | They propose a new type of concept relatedness dataset, WORD, which is annotated by a human . they use this dataset to assess relatedness between Wikipedia concepts using supervised methods. |
| Outcome: | The proposed dataset outperforms existing methods for measuring relatedness between Wikipedia concepts. |
Will it Merge? On The Causes of Model Mergeability (2026.findings-acl)
Copied to clipboard
| Challenge: | Model merging has emerged as a promising technique for combining fine-tuned models into a single expert model without retraining. |
| Approach: | They propose a model merging technique that preserves weak model knowledge . they define mergeability as a property of model updates that captures how well they retain trained knowledge when merged with other model updates. |
| Outcome: | The proposed method preserves weak knowledge in the base model. |
Conversational Document Prediction to Assist Customer Care Agents (2020.emnlp-main)
Copied to clipboard
Jatin Ganhotra, Haggai Roitman, Doron Cohen, Nathaniel Mills, Chulaka Gunasekara, Yosi Mass, Sachindra Joshi, Luis Lastras, David Konopnicki
| Challenge: | Using a conversational search system, the agent/system can ask clarification questions and interactively modify the search results as the conversation progresses. |
| Approach: | They propose to use a public dataset to analyze the task of predicting the documents that customer care agents can use to facilitate users’ needs. |
| Outcome: | The proposed model is more efficient than existing models and is more cost-effective than existing ones. |
More Bang for your Context: Virtual Documents for Question Answering over Long Documents (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Large language models struggle to utilize long contexts efficiently, resulting in a question answering problem. |
| Approach: | They propose a method to generate a short document that contains the most relevant parts for a given context window. |
| Outcome: | The proposed method improves the QA task by providing a short and focused VDoc to the LLM while keeping the context window full. |
Ad-hoc Document Retrieval using Weak-Supervision with BERT and GPT2 (2020.emnlp-main)
Copied to clipboard
| Challenge: | a weakly-supervised method is used for document retrieval tasks . traditional methods are used for ad-hoc querying, but they require large amounts of labeled data . |
| Approach: | They propose a weakly-supervised method for training deep learning models for ad-hoc document retrieval using weak-supervision from the documents in the corpus. |
| Outcome: | The proposed method outperforms state-of-the-art methods on a COVID-19 dataset and two news datasets without the need for labeling data. |
Unsupervised FAQ Retrieval with Question Generation and BERT (2020.acl-main)
Copied to clipboard
| Challenge: | Frequently Asked Questions (FAQ) retrieval requires labeled datasets for training neural models. |
| Approach: | They propose to exploit FAQ pairs to train two BERT models that match user queries to FAQ answers and questions. |
| Outcome: | The proposed model outperforms supervised models on existing datasets and is on par with existing dataset. |
A Summarization System for Scientific Documents (D19-3)
Copied to clipboard
Shai Erera, Michal Shmueli-Scheuer, Guy Feigenblat, Ora Peled Nakash, Odellia Boni, Haggai Roitman, Doron Cohen, Bar Weiner, Yosi Mass, Or Rivlin, Guy Lev, Achiya Jerbi, Jonathan Herzig, Yufang Hou, Charles Jochim, Martin Gleize, Francesca Bonin, Francesca Bonin, David Konopnicki
| Challenge: | a qualitative user study identified the most valuable scenarios for scientific content consumption. |
| Approach: | They propose a system that retrieves and summarizes scientific documents for a given information need. |
| Outcome: | The proposed system ingested 270,000 scientific papers and validated with human experts. |