Yoav Goldberg · page 2
Papers
- Towards String-To-Tree Neural Machine Translation (2017 · 226 citations)
The authors investigate whether knowledge of target-language syntax helps neural machine translation (NMT). - Break It Down: A Question Understanding Benchmark (2020 · 222 citations)
Complex questions in visual QA, reading comprehension, and database querying often have the same compositional structure. - Same Task, More Tokens: the Impact of Input Length on the Reasoning Performance of Large Language Models (2024 · 217 citations)
LLMs advertise ever-larger context windows, but it is unclear whether a model that solves a task on a short prompt can still solve it when the same task sits inside a longer one. - Morphological Inflection Generation with Hard Monotonic Attention (2017 · 211 citations)
The paper addresses the generation of inflected word forms. The model receives a lemma (for example, the German "hart") and a set of morphosyntactic features, and must produce the required word form. Neural seq2seq models with soft attention perform well on large datasets, but on the small CELEX corpus they fall… - Multilingual Part-of-Speech Tagging with Bidirectional Long Short-Term Memory Models and Auxiliary Loss (2016 · 203 citations)
The paper examines how well bi-LSTM part-of-speech taggers work outside English. It tests many languages, several input granularities, small training sets and noisy labels. The authors use a hierarchical bi-LSTM. A lower sequence bi-LSTM builds a word vector from character or unicode byte embeddings. That vector is… - A Dataset of Syntactic-Ngrams over Time from a Very Large Corpus of English Books (2013 · 192 citations)
Large corpora are widely used to learn word properties from context, but syntactic counts at web scale were hard to obtain, and studies seldom shared a common corpus. - Statistical Parsing of Morphologically Rich Languages (SPMRL) What, How and Whither (2010 · 158 citations)
The paper is the opening overview of the first SPMRL workshop (2010). It asks why statistical parsers that perform well on the English Penn Treebank lose much of their accuracy on morphologically rich languages (MRLs), where word forms carry much of the grammatical information and word order tends to be freer. The… - Including Signed Languages in Natural Language Processing (2021 · 147 citations)
This position paper argues that signed languages, used by up to 70 million deaf people across roughly 200 languages, are largely ignored by NLP. - Aligning Faithful Interpretations with their Social Attribution (2021 · 123 citations)
The paper argues that "faithfulness," the property the NLP community usually demands of model interpretations, is too vague to ensure that explanations are useful. - Provable Limitations of Acquiring Meaning from Ungrounded Form: What Will Future Language Models Understand? (2021 · 89 citations)
Can a system trained only on raw text, with no grounding, ever acquire meaning? Merrill, Goldberg, Schwartz and Smith turn the informal debate started by Bender and Koller (2020) into a precise mathematical question. - Studying the Inductive Biases of RNNs with Synthetic Variations of Natural Languages (2019 · 84 citations)
Comparing how well RNNs learn syntax across languages is hard: real languages differ in several typological properties simultaneously, and models get trained on different corpora. - Understanding Transformer Memorization Recall Through Idioms (2023 · 62 citations)
The paper asks how transformer LMs decide to emit a memorized phrase and where that phrase is retrieved internally. - A Non-Monotonic Arc-Eager Transition System for Dependency Parsing (2013 · 55 citations)
The paper relaxes the two deterministic constraints of the arc-eager transition system, which keep each step consistent with earlier head decisions. - LingMess: Linguistically Informed Multi Expert Scorers for Coreference Resolution (2023 · 54 citations)
Neural coreference models score every mention pair with one shared function, although different pairs need different evidence. - F-coref: Fast, Accurate and Easy to Use Coreference Resolution (2022 · 45 citations)
Coreference resolution groups mentions in a text that point to the same entity. It helps tasks such as information extraction, question answering and summarization, but strong existing tools are hard to use at scale because they need a lot of GPU memory and run slowly. The authors release fastcoref, a pip-installable… - Word Segmentation, Unknown-word Resolution, and Morphological Agreement in a Hebrew Parsing System (2013 · 39 citations)
The paper builds a constituency parser for Modern Hebrew on the BerkeleyParser (PCFG with latent annotations) and extends it for a small treebank of 6,220 sentences. - Easy-First Dependency Parsing of Modern Hebrew (2010 · 38 citations)
The authors apply their easy-first, non-directional dependency parser to the Hebrew dependency treebank, a small treebank of a language with rich morphology and free word order. - Where's My Head? Definition, Data Set, and Models for Numeric Fused-Head Identification and Resolution (2019 · 37 citations)
Numbers often appear without the noun they quantify. In "I'm 42" the noun is an age, and in "three at the most" it is a noun mentioned earlier. Humans recover these missing heads easily, but NLP systems mostly ignore them, which causes errors downstream, for example in machine translation. Elazar and Goldberg give… - Joint Hebrew Segmentation and Parsing using a PCFGLA Lattice Parser (2011 · 32 citations)
Hebrew function words are written attached to the next word, so one space-delimited token can have several segmentations, and a pipeline that segments first cannot use the syntax to choose. - Hebrew Dependency Parsing: Initial Results (2009 · 31 citations)
The authors derive a Hebrew dependency treebank (DepTB) from version 2 of the Hebrew constituency treebank. - Text-based NP Enrichment (2022 · 23 citations)
Standard NLP tasks capture only some of the ways entities in a text relate. Semantic role labeling covers links mediated by verbs, and coreference covers identity. Many noun-to-noun connections, whether stated or only implied through prepositions, are left out. The authors propose Text-based NP Enrichment (TNE). For… - Nakdan: Professional Hebrew Diacritizer (2020 · 23 citations)
Nakdan adds vowel points to undiacritized Hebrew, for modern, Rabbinic and poetic texts, through a web interface that lets an editor choose among the alternatives. - Word-Based or Morpheme-Based? Annotation Strategies for Modern Hebrew Clitics (2008 · 23 citations)
The Modern Hebrew Treebank splits pronominal clitics on prepositions, accusative markers and possessive markers into separate leaves: one orthographic word becomes two or more nodes (morpheme-based, MB). - Large Scale Substitution-based Word Sense Induction (2022 · 21 citations)
The paper induces word senses for every frequent word of a large corpus. For each corpus position a masked language model proposes the top five substitutes, and the substitute vectors are clustered per word type into senses. The result is a sense-tagged corpus in which each sense comes with indicative words. Storing… - pyBART: Evidence-based Syntactic Transformations for IE (2020 · 20 citations)
pyBART is an open-source Python library that converts English UD trees into a representation (BART) that makes event structure and many content-word relations explicit, going beyond Enhanced UD.
This page in Markdown → https://md.research.elamur.ai/free/en/authors/yoav-goldberg/page/2/