Many commercial and open-source models claim to detect machine-generated text with extremely high accuracy (99% or more). However, very few of these detectors are evaluated on shared benchmark datasets and even when they are, the datasets used for evaluation are insufficiently challenging—lacking variations in sampling strategy, adversarial attacks, and open-source generative models. In this work we present RAID: the largest and most challenging benchmark dataset for machine-generated text detection. RAID includes over 6 million generations spanning 11 models, 8 domains, 11 adversarial attacks and 4 decoding strategies. Using RAID, we evaluate the out-of-domain and adversarial robustness of 8 open- and 4 closed-source detectors and find that current detectors are easily fooled by adversarial attacks, variations in sampling strategies, repetition penalties, and unseen generative models. We release our data along with a leaderboard to encourage future research.

One type of question that is commonly found in day-to-day scenarios is “fan-out” questions, complex multi-hop, multi-document reasoning questions that require finding information about a large number of entities. However, there exist few resources to evaluate this type of question-answering capability among large language models. To evaluate complex reasoning in LLMs more fully, we present FanOutQA, a high-quality dataset of fan-out question-answer pairs and human-annotated decompositions with English Wikipedia as the knowledge base. We formulate three benchmark settings across our dataset and benchmark 7 LLMs, including GPT-4, LLaMA 2, Claude-2.1, and Mixtral-8x7B, finding that contemporary models still have room to improve reasoning over inter-document dependencies in a long context. We provide our dataset, along with open-source tools to run models to encourage evaluation.

Language model applications are becoming increasingly popular and complex, often including features like tool usage and retrieval augmentation. However, existing frameworks for such applications are often opinionated, deciding for developers how their prompts ought to be formatted and imposing limitations on customizability and reproducibility. To solve this we present Kani: a lightweight, flexible, and model-agnostic open-source framework for building language model applications. Kani helps developers implement a variety of complex features by supporting the core building blocks of chat interaction: model interfacing, chat management, and robust function calling. All Kani core functions are easily overridable and well documented to empower developers to customize functionality for their own needs. Kani thus serves as a useful tool for researchers, hobbyists, and industry professionals alike to accelerate their development while retaining interoperability and fine-grained control.

Researchers in natural language processing have developed large, robust resources for understanding formal Standard American English (SAE), but we lack similar resources for variations of English, such as slang and African American English (AAE). In this work, we use word embeddings and clustering algorithms to group semantically similar words in three datasets, two of which contain high incidence of slang and AAE. Since high-quality clusters would contain related words, we could also infer the meaning of an unfamiliar word based on the meanings of words clustered with it. After clustering, we compute precision and recall scores using WordNet and ConceptNet as gold standards and show that these scores are unimportant when the given resources do not fully represent slang and AAE. Amazon Mechanical Turk and expert evaluations show that clusters with low precision can still be considered high quality, and we propose the new Cluster Split Score as a metric for machine-generated clusters. These contributions emphasize the gap in natural language processing research for variations of English and motivate further work to close it.

Argumentation is a type of discourse where speakers try to persuade their audience about the reasonableness of a claim by presenting supportive arguments. Most work in argument mining has focused on modeling arguments in monologues. We propose a computational model for argument mining in online persuasive discussion forums that brings together the micro-level (argument as product) and macro-level (argument as process) models of argumentation. Fundamentally, this approach relies on identifying relations between components of arguments in a discussion thread. Our approach for relation prediction uses contextual information in terms of fine-tuning a pre-trained language model and leveraging discourse relations based on Rhetorical Structure Theory. We additionally propose a candidate selection method to automatically predict what parts of one’s argument will be targeted by other participants in the discussion. Our models obtain significant improvements compared to recent state-of-the-art approaches using pointer networks and a pre-trained language model.

An idiom is defined as a non-compositional multiword expression, one whose meaning cannot be deduced from the definitions of the component words. This definition does not explicitly define the compositionality of an idiom’s sentiment; this paper aims to determine whether the sentiment of the component words of an idiom is related to the sentiment of that idiom. We use the Dictionary of Affect in Language augmented by WordNet to give each idiom in the Sentiment Lexicon of IDiomatic Expressions (SLIDE) a component-wise sentiment score and compare it to the phrase-level sentiment label crowdsourced by the creators of SLIDE. We find that there is no discernible relation between these two measures of idiom sentiment. This supports the hypothesis that idioms are not compositional for sentiment along with semantics and motivates further work in handling idioms for sentiment analysis.

Argumentative text has been analyzed both theoretically and computationally in terms of argumentative structure that consists of argument components (e.g., claims, premises) and their argumentative relations (e.g., support, attack). Less emphasis has been placed on analyzing the semantic types of argument components. We propose a two-tiered annotation scheme to label claims and premises and their semantic types in an online persuasive forum, Change My View, with the long-term goal of understanding what makes a message persuasive. Premises are annotated with the three types of persuasive modes: ethos, logos, pathos, while claims are labeled as interpretation, evaluation, agreement, or disagreement, the latter two designed to account for the dialogical nature of our corpus. We aim to answer three questions: 1) can humans reliably annotate the semantic types of argument components? 2) are types of premises/claims positioned in recurrent orders? and 3) are certain types of claims and/or premises more likely to appear in persuasive messages than in non-persuasive messages?