The perceived toxicity of language can vary based on someone's identity and beliefs, but this variation is often ignored when collecting toxic language datasets, resulting in dataset and model biases. We seek to understand the who, why, and what behind biases in toxicity annotations. In two online studies with demographically and politically diverse participants, we investigate the effect of annotator identities (who) and beliefs (why), drawing from social psychology research about hate speech, free speech, racist beliefs, political leaning, and more. We disentangle what is annotated as toxic by considering posts with three characteristics: anti-Black language, African American English (AAE) dialect, and vulgarity. Our results show strong associations between annotator identity and beliefs and their ratings of toxicity. Notably, more conservative annotators and those who scored highly on our scale for racist beliefs were less likely to rate anti-Black language as toxic, but more likely to rate AAE as toxic. We additionally present a case study illustrating how a popular toxicity detection system's ratings inherently reflect only specific beliefs and perspectives. Our findings call for contextualizing toxicity labels in social variables, which raises immense implications for toxic language annotation and detection.
相關內容
Social media data such as Twitter messages ("tweets") pose a particular challenge to NLP systems because of their short, noisy, and colloquial nature. Tasks such as Named Entity Recognition (NER) and syntactic parsing require highly domain-matched training data for good performance. While there are some publicly available annotated datasets of tweets, they are all purpose-built for solving one task at a time. As yet there is no complete training corpus for both syntactic analysis (e.g., part of speech tagging, dependency parsing) and NER of tweets. In this study, we aim to create Tweebank-NER, an NER corpus based on Tweebank V2 (TB2), and we use these datasets to train state-of-the-art NLP models. We first annotate named entities in TB2 using Amazon Mechanical Turk and measure the quality of our annotations. We train a Stanza NER model on the new benchmark, achieving competitive performance against other non-transformer NER systems. Finally, we train other Twitter NLP models (a tokenizer, lemmatizer, part of speech tagger, and dependency parser) on TB2 based on Stanza, and achieve state-of-the-art or competitive performance on these tasks. We release the dataset and make the models available to use in an "off-the-shelf" manner for future Tweet NLP research. Our source code, data, and pre-trained models are available at: \url{//github.com/social-machines/TweebankNLP}.
We introduce a generic, language-independent method to collect a large percentage of offensive and hate tweets regardless of their topics or genres. We harness the extralinguistic information embedded in the emojis to collect a large number of offensive tweets. We apply the proposed method on Arabic tweets and compare it with English tweets -- analyzing some cultural differences. We observed a constant usage of these emojis to represent offensiveness in throughout different timelines in Twitter. We manually annotate and publicly release the largest Arabic dataset for offensive, fine-grained hate speech, vulgar and violence content. Furthermore, we benchmark the dataset for detecting offense and hate speech using different transformer architectures and performed in-depth linguistic analysis. We evaluate our models on external datasets -- a Twitter dataset collected using a completely different method, and a multi-platform dataset containing comments from Twitter, YouTube and Facebook, for assessing generalization capability. Competitive results on these datasets suggest that the data collected using our method captures universal characteristics of offensive language. Our findings also highlight the common words used in offensive communications; common targets for hate speech; specific patterns in violence tweets and pinpoints common classification errors due to the need to understand the context, consider culture and background and the presence of sarcasm among others.
Machine learning research typically starts with a fixed data set created early in the process. The focus of the experiments is finding a model and training procedure that result in the best possible performance in terms of some selected evaluation metric. This paper explores how changes in a data set influence the measured performance of a model. Using three publicly available data sets from the legal domain, we investigate how changes to their size, the train/test splits, and the human labelling accuracy impact the performance of a trained deep learning classifier. We assess the overall performance (weighted average) as well as the per-class performance. The observed effects are surprisingly pronounced, especially when the per-class performance is considered. We investigate how "semantic homogeneity" of a class, i.e., the proximity of sentences in a semantic embedding space, influences the difficulty of its classification. The presented results have far reaching implications for efforts related to data collection and curation in the field of AI & Law. The results also indicate that enhancements to a data set could be considered, alongside the advancement of the ML models, as an additional path for increasing classification performance on various tasks in AI & Law. Finally, we discuss the need for an established methodology to assess the potential effects of data set properties.
Offensive language detection and prevention becomes increasing critical for maintaining a healthy social platform and the safe deployment of language models. Despite plentiful researches on toxic and offensive language problem in NLP, existing studies mainly focus on English, while few researches involve Chinese due to the limitation of resources. To facilitate Chinese offensive language detection and model evaluation, we collect COLDataset, a Chinese offensive language dataset containing 37k annotated sentences. With this high-quality dataset, we provide a strong baseline classifier, COLDetector, with 81% accuracy for offensive language detection. Furthermore, we also utilize the proposed \textsc{COLDetector} to study output offensiveness of popular Chinese language models (CDialGPT and CPM). We find that (1) CPM tends to generate more offensive output than CDialGPT, and (2) certain type of prompts, like anti-bias sentences, can trigger offensive outputs more easily.Altogether, our resources and analyses are intended to help detoxify the Chinese online communities and evaluate the safety performance of generative language models. Disclaimer: The paper contains example data that may be considered profane, vulgar, or offensive.
Entity relationship extraction envisions the automatic generation of semantic data models from collections of text, by automatic recognition of entities, by association of entities to form relationships, and by classifying these instances to assign them to entity sets (or classes) and relationship sets (or associations). As a first step in this direction, the Lokahi prototype can extract entities based on the TF*IDF measure, and generate semantic relationships based on document-level co-occurrence statistics, for example with likelihood ratios and pointwise mutual information. This paper presents results of an explorative, prototypical, qualitative and synthetic research, summarizes insights from two research projects and, based on this, indicates an outline for further research in the field of entity relationship extraction from text.
Conventional entity typing approaches are based on independent classification paradigms, which make them difficult to recognize inter-dependent, long-tailed and fine-grained entity types. In this paper, we argue that the implicitly entailed extrinsic and intrinsic dependencies between labels can provide critical knowledge to tackle the above challenges. To this end, we propose \emph{Label Reasoning Network(LRN)}, which sequentially reasons fine-grained entity labels by discovering and exploiting label dependencies knowledge entailed in the data. Specifically, LRN utilizes an auto-regressive network to conduct deductive reasoning and a bipartite attribute graph to conduct inductive reasoning between labels, which can effectively model, learn and reason complex label dependencies in a sequence-to-set, end-to-end manner. Experiments show that LRN achieves the state-of-the-art performance on standard ultra fine-grained entity typing benchmarks, and can also resolve the long tail label problem effectively.
In many scenarios, named entity recognition (NER) models severely suffer from unlabeled entity problem, where the entities of a sentence may not be fully annotated. Through empirical studies performed on synthetic datasets, we find two causes of the performance degradation. One is the reduction of annotated entities and the other is treating unlabeled entities as negative instances. The first cause has less impact than the second one and can be mitigated by adopting pretraining language models. The second cause seriously misguides a model in training and greatly affects its performances. Based on the above observations, we propose a general approach that is capable of eliminating the misguidance brought by unlabeled entities. The core idea is using negative sampling to keep the probability of training with unlabeled entities at a very low level. Experiments on synthetic datasets and real-world datasets show that our model is robust to unlabeled entity problem and surpasses prior baselines. On well-annotated datasets, our model is competitive with state-of-the-art method.
The package cleanNLP provides a set of fast tools for converting a textual corpus into a set of normalized tables. The underlying natural language processing pipeline utilizes Stanford's CoreNLP library, exposing a number of annotation tasks for text written in English, French, German, and Spanish. Annotators include tokenization, part of speech tagging, named entity recognition, entity linking, sentiment analysis, dependency parsing, coreference resolution, and information extraction.
The aim of image captioning is to generate similar captions by machine as human do to describe image contents. Despite many efforts, generating discriminative captions for images remains non-trivial. Most traditional approaches imitate the language structure patterns, thus tend to fall into a stereotype of replicating frequent phrases or sentences and neglect unique aspects of each image. In this work, we propose an image captioning framework with a self-retrieval module as training guidance, which encourages generating discriminative captions. It brings unique advantages: (1) the self-retrieval guidance can act as a metric and an evaluator of caption discriminativeness to assure the quality of generated captions. (2) The correspondence between generated captions and images are naturally incorporated in the generation process without human annotations, and hence our approach could utilize a large amount of unlabeled images to boost captioning performance with no additional laborious annotations. We demonstrate the effectiveness of the proposed retrieval-guided method on MS-COCO and Flickr30k captioning datasets, and show its superior captioning performance with more discriminative captions.
The goal in the NER task is to classify proper nouns of a text into classes such as person, location, and organization. This is an important preprocessing step in many NLP tasks such as question-answering and summarization. Although many research studies have been conducted in this area in English and the state-of-the-art NER systems have reached performances of higher than 90 percent in terms of F1 measure, there are very few research studies for this task in Persian. One of the main important causes of this may be the lack of a standard Persian NER dataset to train and test NER systems. In this research we create a standard, big-enough tagged Persian NER dataset which will be distributed for free for research purposes. In order to construct such a standard dataset, we studied standard NER datasets which are constructed for English researches and found out that almost all of these datasets are constructed using news texts. So we collected documents from ten news websites. Later, in order to provide annotators with some guidelines to tag these documents, after studying guidelines used for constructing CoNLL and MUC standard English datasets, we set our own guidelines considering the Persian linguistic rules.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191