Recently, there has been a strong interest in developing natural language applications that live on personal devices such as mobile phones, watches and IoT with the objective to preserve user privacy and have low memory. Advances in Locality-Sensitive Hashing (LSH)-based projection networks have demonstrated state-of-the-art performance in various classification tasks without explicit word (or word-piece) embedding lookup tables by computing on-the-fly text representations. In this paper, we show that the projection based neural classifiers are inherently robust to misspellings and perturbations of the input text. We empirically demonstrate that the LSH projection based classifiers are more robust to common misspellings compared to BiLSTMs (with both word-piece & word-only tokenization) and fine-tuned BERT based methods. When subject to misspelling attacks, LSH projection based classifiers had a small average accuracy drop of 2.94% across multiple classifications tasks, while the fine-tuned BERT model accuracy had a significant drop of 11.44%.
In this paper we present a method to learn word embeddings that are resilient to misspellings. Existing word embeddings have limited applicability to malformed texts, which contain a non-negligible amount of out-of-vocabulary words. We propose a method combining FastText with subwords and a supervised task of learning misspelling patterns. In our method, misspellings of each word are embedded close to their correct variants. We train these embeddings on a new dataset we are releasing publicly. Finally, we experimentally show the advantages of this approach on both intrinsic and extrinsic NLP t…
Georgetown University Medical Center, Georgetown University（乔治城大学医学中心，乔治城大学）；Tippie College of Business, University of Iowa（蒂皮商学院，爱荷华大学）
The focus of our paper is the identification and correction of non-word errors in OCR text. Such errors may be the result of incorrect insertion, deletion, or substitution of a character, or the transposition of two adjacent characters within a single word. Or, it can be the result of word boundary problems that lead to run-on errors and incorrect-split errors. The traditional N-gram correction methods can handle single-word errors effectively. However, they show limitations when dealing with split and merge errors. In this paper, we develop an unsupervised method that can handle both errors. …
Spelling irregularities, known now as spelling mistakes, have been found for several centuries. As humans, we are able to understand most of the misspelled words based on their location in the sentence, perceived pronunciation, and context. Unlike humans, computer systems do not possess the convenient auto complete functionality of which human brains are capable. While many programs provide spelling correction functionality, many systems do not take context into account. Moreover, Artificial Intelligence systems function in the way they are trained on. With many current Natural Language Proces…
University of Illinois at Chicago（伊利诺伊大学芝加哥分校）；Salesforce Research（Salesforce研究院）
There is an increasing amount of literature that claims the brittleness of deep neural networks in dealing with adversarial examples that are created maliciously. It is unclear, however, how the models will perform in realistic scenarios where \textit{natural rather than malicious} adversarial instances often exist. This work systematically explores the robustness of BERT, the state-of-the-art Transformer-style model in NLP, in dealing with noisy data, particularly mistakes in typing the keyboard, that occur inadvertently. Intensive experiments on sentiment analysis and question answering benc…
Stanford University（斯坦福大学）；University of Edinburgh（爱丁堡大学）；University of Washington（华盛顿大学）；Harvard University（哈佛大学）
Intuitively, human readers cope easily with errors in text; typos, misspelling, word substitutions, etc. do not unduly disrupt natural reading. Previous work indicates that letter transpositions result in increased reading times, but it is unclear if this effect generalizes to more natural errors. In this paper, we report an eye-tracking study that compares two error types (letter transpositions and naturally occurring misspelling) and two error rates (10% or 50% of all words contain errors). We find that human readers show unimpaired comprehension in spite of these errors, but error words cau…
Lebanese Association for Computational Sciences（黎巴嫩计算科学协会）
Spell-checking is the process of detecting and sometimes providing suggestions for incorrectly spelled words in a text. Basically, the larger the dictionary of a spell-checker is, the higher is the error detection rate; otherwise, misspellings would pass undetected. Unfortunately, traditional dictionaries suffer from out-of-vocabulary and data sparseness problems as they do not encompass large vocabulary of words indispensable to cover proper names, domain-specific terms, technical jargons, special acronyms, and terminologies. As a result, spell-checkers will incur low error detection and corr…
Mila, Université de Montréal（蒙特利尔大学米拉研究所）；Google Research（谷歌研究院）；Google（谷歌）
Neural word representations are at the core of many state-of-the-art natural language processing models. A widely used approach is to pre-train, store and look up word or character embedding matrices. While useful, such representations occupy huge memory making it hard to deploy on-device and often do not generalize to unknown words due to vocabulary pruning. In this paper, we propose a skip-gram based architecture coupled with Locality-Sensitive Hashing (LSH) projections to learn efficient dynamically computable representations. Our model does not need to store lookup tables as representation…
We present an unsupervised context-sensitive spelling correction method for clinical free-text that uses word and character n-gram embeddings. Our method generates misspelling replacement candidates and ranks them according to their semantic fit, by calculating a weighted cosine similarity between the vectorized representation of a candidate and the misspelling context. To tune the parameters of this model, we generate self-induced spelling error corpora. We perform our experiments for two languages. For English, we greatly outperform off-the-shelf spelling correction tools on a manually annot…
Tokyo Institute of Technology（东京工业大学）；National Institute of Advanced Industrial Science and Technology（国立先进工业科学技术研究所）
We propose neural models that can normalize text by considering the similarities of word strings and sounds. We experimentally compared a model that considers the similarities of both word strings and sounds, a model that considers only the similarity of word strings or of sounds, and a model without the similarities as a baseline. Results showed that leveraging the word string similarity succeeded in dealing with misspellings and abbreviations, and taking into account the sound similarity succeeded in dealing with phonetic substitutions and emphasized characters. So that the proposed models a…
Machine learning based language models have recently made significant progress, which introduces a danger to spread misinformation. To combat this potential danger, several methods have been proposed for detecting text written by these language models. This paper presents two classes of black-box attacks on these detectors, one which randomly replaces characters with homoglyphs, and the other a simple scheme to purposefully misspell words. The homoglyph and misspelling attacks decrease a popular neural text detector's recall on neural text from 97.44% to 0.26% and 22.68%, respectively. Results…
Lebanese Association for Computational Sciences (LACSC)（黎巴嫩计算科学协会）
In computing, spell checking is the process of detecting and sometimes providing spelling suggestions for incorrectly spelled words in a text. Basically, a spell checker is a computer program that uses a dictionary of words to perform spell checking. The bigger the dictionary is, the higher is the error detection rate. The fact that spell checkers are based on regular dictionaries, they suffer from data sparseness problem as they cannot capture large vocabulary of words including proper names, domain-specific terms, technical jargons, special acronyms, and terminologies. As a result, they exhi…
Recently, generating adversarial examples has become an important means of measuring robustness of a deep learning model. Adversarial examples help us identify the susceptibilities of the model and further counter those vulnerabilities by applying adversarial training techniques. In natural language domain, small perturbations in the form of misspellings or paraphrases can drastically change the semantics of the text. We propose a reinforcement learning based approach towards generating adversarial examples in black-box settings. We demonstrate that our method is able to fool well-trained mode…
Aalborg Universitet（奥尔堡大学）；Insight Centre for Data Analytics（数据分析洞察中心）；University College Dublin（都柏林大学学院）
An important task for recommender system is to generate explanations according to a user's preferences. Most of the current methods for explainable recommendations use structured sentences to provide descriptions along with the recommendations they produce. However, those methods have neglected the review-oriented way of writing a text, even though it is known that these reviews have a strong influence over user's decision. In this paper, we propose a method for the automatic generation of natural language explanations, for predicting how a user would write about an item, based on user ratings…
This thesis addresses automatic lexical error recovery and tokenization of corrupt text input. We propose a technique that can automatically correct misspellings, segmentation errors and real-word errors in a unified framework that uses both a model of language production and a model of the typing behavior, and which makes tokenization part of the recovery process. The typing process is modeled as a noisy channel where Hidden Markov Models are used to model the channel characteristics. Weak statistical language models are used to predict what sentences are likely to be transmitted through the …
School of Computer Science and Technology, Huazhong University of Science and Technology（华中科技大学计算机科学与技术学院）；Computer Science Department, University of California, Los Angeles（加州大学洛杉矶分校计算机科学系）
Adversarial training is the most empirically successful approach in improving the robustness of deep neural networks for image classification.For text classification, however, existing synonym substitution based adversarial attacks are effective but not efficient to be incorporated into practical text adversarial training. Gradient-based attacks, which are very efficient for images, are hard to be implemented for synonym substitution based text attacks due to the lexical, grammatical and semantic constraints and the discrete text input space. Thereby, we propose a fast text adversarial attack …
We apply entropy agglomeration (EA), a recently introduced algorithm, to cluster the words of a literary text. EA is a greedy agglomerative procedure that minimizes projection entropy (PE), a function that can quantify the segmentedness of an element set. To apply it, the text is reduced to a feature allocation, a combinatorial object to represent the word occurences in the text's paragraphs. The experiment results demonstrate that EA, despite its reduction and simplicity, is useful in capturing significant relationships among the words in the text. This procedure was implemented in Python and…
International Institute of Information Technology, Hyderabad（海得拉巴国际信息技术学院）
Sentiment analysis (SA) using code-mixed data from social media has several applications in opinion mining ranging from customer satisfaction to social campaign analysis in multilingual societies. Advances in this area are impeded by the lack of a suitable annotated dataset. We introduce a Hindi-English (Hi-En) code-mixed dataset for sentiment analysis and perform empirical analysis comparing the suitability and performance of various state-of-the-art SA methods in social media. In this paper, we introduce learning sub-word level representations in LSTM (Subword-LSTM) architecture instead of c…
Social networks have become an indispensable part of our lives, with billions of people producing ever-increasing amounts of text. At such scales, content policies and their enforcement become paramount. To automate moderation, questionable content is detected by Natural Language Processing (NLP) classifiers. However, high-performance classifiers are hampered by misspellings and adversarial text perturbations. In this paper, we classify intentional and unintentional adversarial text perturbation into ten types and propose a deobfuscation pipeline to make NLP models robust to such perturbations…
Recently, deep learning based natural language processing techniques are being extensively used to deal with spam mail, censorship evaluation in social networks, among others. However, there is only a couple of works evaluating the vulnerabilities of such deep neural networks. Here, we go beyond attacks to investigate, for the first time, universal rules, i.e., rules that are sample agnostic and therefore could turn any text sample in an adversarial one. In fact, the universal rules do not use any information from the method itself (no information from the method, gradient information or train…
LACSC – Lebanese Association for Computational Sciences（黎巴嫩计算科学协会（LACSC））
With the advent of digital optical scanners, a lot of paper-based books, textbooks, magazines, articles, and documents are being transformed into an electronic version that can be manipulated by a computer. For this purpose, OCR, short for Optical Character Recognition was developed to translate scanned graphical text into editable computer text. Unfortunately, OCR is still imperfect as it occasionally mis-recognizes letters and falsely identifies scanned text, leading to misspellings and linguistics errors in the OCR output text. This paper proposes a post-processing context-based error corre…
Mila, Université de Montréal（蒙特利尔大学米拉研究所）；Amazon Alexa（亚马逊Alexa）；Google（谷歌）
At the heart of text based neural models lay word representations, which are powerful but occupy a lot of memory making it challenging to deploy to devices with memory constraints such as mobile phones, watches and IoT. To surmount these challenges, we introduce ProFormer -- a projection based transformer architecture that is faster and lighter making it suitable to deploy to memory constraint devices and preserve user privacy. We use LSH projection layer to dynamically generate word representations on-the-fly without embedding lookup tables leading to significant memory footprint reduction fr…
Knowledge representation is an important, long-history topic in AI, and there have been a large amount of work for knowledge graph embedding which projects symbolic entities and relations into low-dimensional, real-valued vector space. However, most embedding methods merely concentrate on data fitting and ignore the explicit semantic expression, leading to uninterpretable representations. Thus, traditional embedding methods have limited potentials for many applications such as question answering, and entity classification. To this end, this paper proposes a semantic representation method for k…
Wangxuan Institute of Computer Technology, Peking University（北京大学王选计算机技术研究所）；Tencent AI Lab（腾讯人工智能实验室）；R&D Center Singapore, Machine Intelligence Technology, Alibaba DAMO Academy（阿里巴巴达摩院机器智能技术新加坡研发中心）；Center for Data Science, China（中国数据科学中心）
Text style transfer task requires the model to transfer a sentence of one style to another style while retaining its original content meaning, which is a challenging problem that has long suffered from the shortage of parallel data. In this paper, we first propose a semi-supervised text style transfer model that combines the small-scale parallel data with the large-scale nonparallel data. With these two types of training data, we introduce a projection function between the latent space of different styles and design two constraints to train it. We also introduce two other simple but effective …
A public dataset, with a variety of properties suitable for sentiment analysis [1], event prediction, trend detection and other text mining applications, is needed in order to be able to successfully perform analysis studies. The vast majority of data on social media is text-based and it is not possible to directly apply machine learning processes into these raw data, since several different processes are required to prepare the data before the implementation of the algorithms. For example, different misspellings of same word enlarge the word vector space unnecessarily, thereby it leads to red…
Cross-Lingual Clinical Annotation Projection as Constrained Text Generation: A Six-Language Study
Universidad de Málaga（马拉加大学）；Research Institute of Multilingual Language Technologies, Universidad de Málaga（马拉加大学多语言语言技术研究所）；IBIMA Plataforma BIONAND, Instituto de Investigación Biomédica de Málaga（IBIMA BIONAND平台，马拉加生物医学研究所）
iFLYTEK Research（科大讯飞研究）；State Key Laboratory of Cognitive Intelligence（认知智能国家重点实验室）；Harbin Institute of Technology（哈尔滨工业大学）；Research Center for Social Computing and Information Retrieval (SCIR)（社会计算与信息检索研究中心（SCIR））；iFLYTEK AI Research (Hebei)（科大讯飞AI研究（河北））；University of Maryland（马里兰大学）
Most pre-trained language models (PLMs) construct word representations at subword level with Byte-Pair Encoding (BPE) or its variations, by which OOV (out-of-vocab) words are almost avoidable. However, those methods split a word into subword units and make the representation incomplete and fragile. In this paper, we propose a character-aware pre-trained language model named CharBERT improving on the previous methods (such as BERT, RoBERTa) to tackle these problems. We first construct the contextual word embedding for each token from the sequential character representations, then fuse the repre…
Shanghai Jiao Tong University（上海交通大学）；State Key Lab of Media Convergence Production Technology and Systems（媒体融合生产技术与系统国家重点实验室）
Given a database schema, Text-to-SQL aims to translate a natural language question into the corresponding SQL query. Under the setup of cross-domain, traditional semantic parsing models struggle to adapt to unseen database schemas. To improve the model generalization capability for rare and unseen schemas, we propose a new architecture, ShadowGNN, which processes schemas at abstract and semantic levels. By ignoring names of semantic items in databases, abstract schemas are exploited in a well-designed graph projection neural network to obtain delexicalized representation of question and schema…
Part of speech (POS) tagging is a familiar NLP task. State of the art taggers routinely achieve token-level accuracies of over 97% on news body text, evidence that the problem is well understood. However, the register of English news headlines, "headlinese", is very different from the register of long-form text, causing POS tagging models to underperform on headlines. In this work, we automatically annotate news headlines with POS tags by projecting predicted tags from corresponding sentences in news bodies. We train a multi-domain POS tagger on both long-form and headline text and show that j…
