id
stringlengths
33
34
title
stringlengths
6
218
authors
stringlengths
5
6.98k
categories
stringlengths
5
188
summary
stringlengths
34
3.26k
link_pdf
stringlengths
33
34
updated
timestamp[s]
published
timestamp[s]
embeddings
dict
http://arxiv.org/abs/2302.03905v1
COMBO: A Complete Benchmark for Open KG Canonicalization
Chengyue Jiang, Yong Jiang, Weiqi Wu, Yuting Zheng, Pengjun Xie, Kewei Tu
cs.CL, cs.AI
Open knowledge graph (KG) consists of (subject, relation, object) triples extracted from millions of raw text. The subject and object noun phrases and the relation in open KG have severe redundancy and ambiguity and need to be canonicalized. Existing datasets for open KG canonicalization only provide gold entity-level ...
http://arxiv.org/pdf/2302.03905v1
2023-02-08T06:46:01
2023-02-08T06:46:01
{ "title": [ -0.071777344, -0.012588501, -0.05050659, -0.08660889, -0.0026683807, 0.02998352, -0.09094238, -0.011375427, -0.003326416, 0.062316895, 0.039520264, 0.0011425018, 0.029296875, -0.041778564, -0.03488159, 0.033416748, 0.010902405, 0.012...
http://arxiv.org/abs/2302.03896v3
EvoText: Enhancing Natural Language Generation Models via Self-Escalation Learning for Up-to-Date Knowledge and Improved Performance
Zhengqing Yuan, Huiwen Xue, Chao Zhang, Yongming Liu
cs.CL
In recent years, pretrained models have been widely used in various fields, including natural language understanding, computer vision, and natural language generation. However, the performance of these language generation models is highly dependent on the model size and the dataset size. While larger models excel in so...
http://arxiv.org/pdf/2302.03896v3
2023-04-13T07:29:20
2023-02-08T06:09:55
{ "title": [ -0.013145447, -0.01109314, -0.055725098, -0.062927246, 0.002779007, -0.0090408325, -0.04168701, 0.012626648, 0.0011940002, 0.018661499, -0.026916504, 0.009735107, -0.0066108704, -0.014770508, -0.0309906, 0.003780365, -0.008049011, 0....
http://arxiv.org/abs/2302.03848v1
Controlling Personality Style in Dialogue with Zero-Shot Prompt-Based Learning
Angela Ramirez, Mamon Alsalihy, Kartik Aggarwal, Cecilia Li, Liren Wu, Marilyn Walker
cs.CL
Prompt-based or in-context learning has achieved high zero-shot performance on many natural language generation (NLG) tasks. Here we explore the performance of prompt-based learning for simultaneously controlling the personality and the semantic accuracy of an NLG for task-oriented dialogue. We experiment with prompt-b...
http://arxiv.org/pdf/2302.03848v1
2023-02-08T02:45:21
2023-02-08T02:45:21
{ "title": [ -0.0034065247, -0.021591187, -0.045288086, -0.035858154, -0.014190674, -0.027801514, -0.004257202, 0.018310547, -0.007713318, 0.033996582, 0.0070266724, 0.00907135, 0.0028457642, -0.038726807, -0.01979065, 0.012229919, -0.008659363, ...
http://arxiv.org/abs/2302.03822v2
Clinical BioBERT Hyperparameter Optimization using Genetic Algorithm
Navya Martin Kollapally, James Geller
cs.AI, cs.CL, cs.LG
Clinical factors account only for a small portion, about 10-30%, of the controllable factors that affect an individual's health outcomes. The remaining factors include where a person was born and raised, where he/she pursued their education, what their work and family environment is like, etc. These factors are collect...
http://arxiv.org/pdf/2302.03822v2
2023-02-13T03:55:10
2023-02-08T01:11:59
{ "title": [ -0.020492554, 0.0066108704, -0.021224976, -0.06781006, -0.00944519, 0.0155181885, -0.07434082, -0.018310547, -0.010826111, 0.053955078, 0.024719238, 0.0069465637, 0.0062217712, -0.044921875, -0.03363037, 0.020568848, 0.004840851, 0.0...
http://arxiv.org/abs/2302.03815v1
Long Text and Multi-Table Summarization: Dataset and Method
Shuaiqi Liu, Jiannong Cao, Ruosong Yang, Zhiyuan Wen
cs.CL, cs.AI, I.2.7; I.7
Automatic document summarization aims to produce a concise summary covering the input document's salient information. Within a report document, the salient information can be scattered in the textual and non-textual content. However, existing document summarization datasets and methods usually focus on the text and fil...
http://arxiv.org/pdf/2302.03815v1
2023-02-08T00:46:55
2023-02-08T00:46:55
{ "title": [ -0.010475159, 0.012840271, -0.056610107, -0.063964844, 0.020385742, -0.031082153, -0.04171753, -0.008491516, -0.019958496, 0.051849365, -0.02532959, 0.024414062, -0.004360199, -0.010856628, -0.03186035, -0.0036525726, 0.0073394775, 0...
http://arxiv.org/abs/2302.03780v3
Reliable Natural Language Understanding with Large Language Models and Answer Set Programming
Abhiramon Rajasekharan, Yankai Zeng, Parth Padalkar, Gopal Gupta
cs.CL, cs.AI
Humans understand language by extracting information (meaning) from sentences, combining it with existing commonsense knowledge, and then performing reasoning to draw conclusions. While large language models (LLMs) such as GPT-3 and ChatGPT are able to leverage patterns in the text to solve a variety of NLP tasks, they...
http://arxiv.org/pdf/2302.03780v3
2023-08-30T08:56:37
2023-02-07T22:37:21
{ "title": [ -0.037231445, -0.0129776, -0.03149414, -0.07897949, 0.0026187897, 0.023086548, -0.06451416, 0.0082092285, 0.00894165, 0.007583618, -0.020263672, -0.010688782, -0.017990112, -0.018539429, -0.04397583, -0.0035305023, -0.018051147, 0.01...
http://arxiv.org/abs/2302.03773v1
What Matters In The Structured Pruning of Generative Language Models?
Michael Santacroce, Zixin Wen, Yelong Shen, Yuanzhi Li
cs.CL, cs.LG
Auto-regressive large language models such as GPT-3 require enormous computational resources to use. Traditionally, structured pruning methods are employed to reduce resource usage. However, their application to and efficacy for generative language models is heavily under-explored. In this paper we conduct an comprehen...
http://arxiv.org/pdf/2302.03773v1
2023-02-07T22:05:55
2023-02-07T22:05:55
{ "title": [ -0.023880005, -0.028900146, -0.018829346, -0.060028076, -0.058746338, 0.0088272095, -0.09051514, 0.027557373, 0.005645752, 0.019927979, -0.02394104, 0.011100769, 0.0017089844, -0.035339355, 0.0012197495, -0.000040113926, -0.0017623901, ...
http://arxiv.org/abs/2302.03765v1
Transformer-based Models for Long-Form Document Matching: Challenges and Empirical Analysis
Akshita Jha, Adithya Samavedhi, Vineeth Rakesh, Jaideep Chandrashekar, Chandan K. Reddy
cs.CL
Recent advances in the area of long document matching have primarily focused on using transformer-based models for long document encoding and matching. There are two primary challenges associated with these models. Firstly, the performance gain provided by transformer-based models comes at a steep cost - both in terms ...
http://arxiv.org/pdf/2302.03765v1
2023-02-07T21:51:05
2023-02-07T21:51:05
{ "title": [ -0.012969971, 0.023590088, -0.031051636, -0.04144287, 0.02154541, -0.0036849976, -0.041412354, -0.03567505, 0.0071372986, 0.035614014, -0.0317688, -0.0078125, 0.043182373, -0.009490967, -0.04006958, -0.013938904, -0.013671875, 0.0501...
http://arxiv.org/abs/2302.03754v1
Augmenting Zero-Shot Dense Retrievers with Plug-in Mixture-of-Memories
Suyu Ge, Chenyan Xiong, Corby Rosset, Arnold Overwijk, Jiawei Han, Paul Bennett
cs.CL
In this paper we improve the zero-shot generalization ability of language models via Mixture-Of-Memory Augmentation (MoMA), a mechanism that retrieves augmentation documents from multiple information corpora ("external memories"), with the option to "plug in" new memory at inference time. We develop a joint learning me...
http://arxiv.org/pdf/2302.03754v1
2023-02-07T20:59:31
2023-02-07T20:59:31
{ "title": [ 0.05303955, -0.023147583, -0.0115356445, -0.074279785, -0.030288696, -0.015930176, -0.008979797, 0.026641846, 0.0022773743, 0.04034424, -0.011985779, 0.014526367, 0.010597229, -0.04107666, -0.00069856644, 0.020187378, 0.014930725, 0....
http://arxiv.org/abs/2302.03693v4
Concept Algebra for (Score-Based) Text-Controlled Generative Models
Zihao Wang, Lin Gui, Jeffrey Negrea, Victor Veitch
cs.CL, cs.LG, stat.ML
This paper concerns the structure of learned representations in text-guided generative models, focusing on score-based models. A key property of such models is that they can compose disparate concepts in a `disentangled' manner. This suggests these models have internal representations that encode concepts in a `disenta...
http://arxiv.org/pdf/2302.03693v4
2023-11-01T03:23:34
2023-02-07T20:43:48
{ "title": [ -0.027770996, -0.029647827, -0.021911621, -0.041046143, -0.0385437, 0.005142212, -0.05911255, -0.00995636, 0.032104492, 0.031402588, -0.014350891, 0.017089844, 0.012542725, -0.032073975, -0.03363037, 0.02180481, -0.022323608, 0.01398...
http://arxiv.org/abs/2302.03716v1
Mining Effective Features Using Quantum Entropy for Humor Recognition
Yang Liu, Yuexian Hou
cs.CL
Humor recognition has been extensively studied with different methods in the past years. However, existing studies on humor recognition do not understand the mechanisms that generate humor. In this paper, inspired by the incongruity theory, any joke can be divided into two components (the setup and the punchline). Both...
http://arxiv.org/pdf/2302.03716v1
2023-02-07T19:09:09
2023-02-07T19:09:09
{ "title": [ 0.019439697, 0.011833191, -0.041046143, -0.021957397, 0.018005371, -0.025527954, -0.0041122437, -0.029708862, 0.010848999, 0.05996704, 0.016860962, -0.044921875, 0.017456055, -0.02947998, -0.10443115, -0.009132385, -0.02722168, 0.068...
http://arxiv.org/abs/2302.04089v2
ZipLM: Inference-Aware Structured Pruning of Language Models
Eldar Kurtic, Elias Frantar, Dan Alistarh
cs.LG, cs.CL
The breakthrough performance of large language models (LLMs) comes with major computational footprints and high deployment costs. In this paper, we progress towards resolving this problem by proposing a novel structured compression approach for LLMs, called ZipLM. ZipLM achieves state-of-the-art accuracy-vs-speedup, wh...
http://arxiv.org/pdf/2302.04089v2
2023-10-26T06:42:40
2023-02-07T18:55:28
{ "title": [ -0.03778076, -0.03692627, -0.035858154, -0.105407715, -0.017318726, 0.034240723, -0.10192871, 0.04989624, 0.00434494, 0.00674057, -0.016326904, 0.0058059692, 0.01637268, -0.020492554, -0.022476196, 0.007019043, -0.017440796, 0.034362...
http://arxiv.org/abs/2302.03675v2
Auditing Gender Presentation Differences in Text-to-Image Models
Yanzhe Zhang, Lu Jiang, Greg Turk, Diyi Yang
cs.CV, cs.AI, cs.CL, cs.CY
Text-to-image models, which can generate high-quality images based on textual input, have recently enabled various content-creation tools. Despite significantly affecting a wide range of downstream applications, the distributions of these generated images are still not fully understood, especially when it comes to the ...
http://arxiv.org/pdf/2302.03675v2
2023-02-08T01:55:54
2023-02-07T18:52:22
{ "title": [ -0.028152466, 0.009666443, -0.03866577, -0.021850586, 0.0044174194, -0.016616821, -0.029052734, -0.0231781, 0.025360107, 0.048797607, -0.030807495, 0.004547119, 0.021469116, -0.002292633, -0.054748535, -0.0005722046, -0.0135269165, 0...
http://arxiv.org/abs/2302.03668v2
Hard Prompts Made Easy: Gradient-Based Discrete Optimization for Prompt Tuning and Discovery
Yuxin Wen, Neel Jain, John Kirchenbauer, Micah Goldblum, Jonas Geiping, Tom Goldstein
cs.LG, cs.CL
The strength of modern generative models lies in their ability to be controlled through text-based prompts. Typical "hard" prompts are made from interpretable words and tokens, and must be hand-crafted by humans. There are also "soft" prompts, which consist of continuous feature vectors. These can be discovered using p...
http://arxiv.org/pdf/2302.03668v2
2023-06-01T12:26:45
2023-02-07T18:40:18
{ "title": [ -0.046295166, -0.017791748, -0.02192688, -0.028961182, -0.030212402, 0.00050354004, -0.05807495, -0.020629883, -0.028335571, 0.018569946, -0.015541077, 0.014770508, -0.017227173, -0.039794922, -0.014472961, -0.0115356445, 0.0026950836, ...
http://arxiv.org/abs/2302.03645v2
Exploitation and exploration in text evolution. Quantifying planning and translation flows during writing
Donald Ruggiero Lo Sardo, Pietro Gravino, Christine Cuskley, Vittorio Loreto
cs.CL
Writing is a complex process at the center of much of modern human activity. Despite it appears to be a linear process, writing conceals many highly non-linear processes. Previous research has focused on three phases of writing: planning, translation and transcription, and revision. While research has shown these are n...
http://arxiv.org/pdf/2302.03645v2
2023-02-08T11:25:23
2023-02-07T17:52:33
{ "title": [ -0.062469482, -0.00712204, -0.0104904175, 0.034942627, -0.044311523, -0.008178711, -0.00072813034, -0.026870728, -0.032714844, 0.033569336, -0.036071777, 0.037506104, 0.018661499, -0.012863159, 0.015182495, -0.024215698, -0.006088257, ...
http://arxiv.org/abs/2302.03589v1
CALaMo: a Constructionist Assessment of Language Models
Ludovica Pannitto, Aurélie Herbelot
cs.CL
This paper presents a novel framework for evaluating Neural Language Models' linguistic abilities using a constructionist approach. Not only is the usage-based model in line with the underlying stochastic philosophy of neural architectures, but it also allows the linguist to keep meaning as a determinant factor in the ...
http://arxiv.org/pdf/2302.03589v1
2023-02-07T16:56:48
2023-02-07T16:56:48
{ "title": [ -0.048431396, -0.009468079, -0.042938232, -0.052978516, 0.0015487671, 0.015296936, -0.060272217, 0.0029964447, 0.0034599304, 0.031051636, -0.009780884, 0.0152282715, -0.012863159, -0.011940002, -0.06021118, 0.011886597, -0.030044556, ...
http://arxiv.org/abs/2302.03694v1
Characterizing Financial Market Coverage using Artificial Intelligence
Jean Marie Tshimula, D'Jeff K. Nkashama, Patrick Owusu, Marc Frappier, Pierre-Martin Tardif, Froduald Kabanza, Armelle Brun, Jean-Marc Patenaude, Shengrui Wang, Belkacem Chikhaoui
q-fin.ST, cs.AI, cs.CL, cs.CY, cs.LG
This paper scrutinizes a database of over 4900 YouTube videos to characterize financial market coverage. Financial market coverage generates a large number of videos. Therefore, watching these videos to derive actionable insights could be challenging and complex. In this paper, we leverage Whisper, a speech-to-text mod...
http://arxiv.org/pdf/2302.03694v1
2023-02-07T16:03:33
2023-02-07T16:03:33
{ "title": [ -0.009979248, 0.01977539, -0.03378296, -0.03173828, 0.0036334991, -0.012619019, 0.0016126633, -0.027862549, -0.02319336, 0.04626465, -0.03579712, -0.001663208, -0.008605957, -0.014701843, -0.045715332, -0.014701843, -0.0069274902, 0....
http://arxiv.org/abs/2302.03528v1
Efficiently Upgrading Multilingual Machine Translation Models to Support More Languages
Simeng Sun, Maha Elbayad, Anna Sun, James Cross
cs.CL
With multilingual machine translation (MMT) models continuing to grow in size and number of supported languages, it is natural to reuse and upgrade existing models to save computation as data becomes available in more languages. However, adding new languages requires updating the vocabulary, which complicates the reuse...
http://arxiv.org/pdf/2302.03528v1
2023-02-07T15:20:13
2023-02-07T15:20:13
{ "title": [ -0.020248413, -0.018188477, -0.023834229, -0.06756592, -0.016189575, 0.005783081, -0.053466797, 0.010353088, -0.007255554, 0.056640625, -0.01663208, -0.0015621185, 0.046295166, -0.02293396, -0.022613525, 0.02947998, 0.0033931732, 0.0...
http://arxiv.org/abs/2302.03512v3
A Survey on Arabic Named Entity Recognition: Past, Recent Advances, and Future Trends
Xiaoye Qu, Yingjie Gu, Qingrong Xia, Zechang Li, Zhefeng Wang, Baoxing Huai
cs.CL
As more and more Arabic texts emerged on the Internet, extracting important information from these Arabic texts is especially useful. As a fundamental technology, Named entity recognition (NER) serves as the core component in information extraction technology, while also playing a critical role in many other Natural La...
http://arxiv.org/pdf/2302.03512v3
2023-08-08T13:27:29
2023-02-07T14:56:52
{ "title": [ -0.029647827, -0.012413025, -0.043304443, -0.08251953, 0.025100708, 0.008140564, -0.029220581, 0.029968262, -0.008293152, 0.017807007, -0.009849548, 0.015655518, -0.012268066, -0.019302368, -0.031311035, -0.0013914108, 0.0005507469, ...
http://arxiv.org/abs/2302.03508v1
Cluster-Level Contrastive Learning for Emotion Recognition in Conversations
Kailai Yang, Tianlin Zhang, Hassan Alhuzali, Sophia Ananiadou
cs.CL, cs.AI
A key challenge for Emotion Recognition in Conversations (ERC) is to distinguish semantically similar emotions. Some works utilise Supervised Contrastive Learning (SCL) which uses categorical emotion labels as supervision signals and contrasts in high-dimensional semantic space. However, categorical labels fail to prov...
http://arxiv.org/pdf/2302.03508v1
2023-02-07T14:49:20
2023-02-07T14:49:20
{ "title": [ 0.016784668, 0.0027618408, -0.018127441, -0.0496521, 0.013954163, -0.00945282, -0.020965576, -0.016204834, 0.033233643, 0.029693604, -0.034118652, -0.0036392212, 0.0026607513, -0.020828247, -0.060424805, 0.000364542, -0.012519836, 0....
http://arxiv.org/abs/2302.03491v1
Learning Translation Quality Evaluation on Low Resource Languages from Large Language Models
Amirkeivan Mohtashami, Mauro Verzetti, Paul K. Rubenstein
cs.CL, cs.LG
Learned metrics such as BLEURT have in recent years become widely employed to evaluate the quality of machine translation systems. Training such metrics requires data which can be expensive and difficult to acquire, particularly for lower-resource languages. We show how knowledge can be distilled from Large Language Mo...
http://arxiv.org/pdf/2302.03491v1
2023-02-07T14:35:35
2023-02-07T14:35:35
{ "title": [ -0.052734375, -0.0048713684, -0.021575928, -0.0335083, -0.010879517, 0.0129776, -0.04852295, -0.0047721863, -0.0055618286, 0.032958984, -0.016586304, 0.00056266785, -0.0077705383, -0.011932373, -0.056396484, -0.0040130615, -0.012382507, ...
http://arxiv.org/abs/2302.03490v1
Natural Language Processing for Policymaking
Zhijing Jin, Rada Mihalcea
cs.CL, cs.AI, cs.CY, cs.LG
Language is the medium for many political activities, from campaigns to news reports. Natural language processing (NLP) uses computational tools to parse text into key information that is needed for policymaking. In this chapter, we introduce common methods of NLP, including text classification, topic modeling, event e...
http://arxiv.org/pdf/2302.03490v1
2023-02-07T14:34:39
2023-02-07T14:34:39
{ "title": [ 0.001578331, -0.013320923, -0.021636963, -0.058288574, -0.00623703, 0.00048279762, -0.025650024, 0.00083971024, -0.040130615, 0.019470215, -0.03970337, 0.0066871643, -0.03604126, -0.01084137, -0.034088135, -0.025222778, -0.011619568, ...
http://arxiv.org/abs/2302.03475v1
Entity-Aware Dual Co-Attention Network for Fake News Detection
Sin-Han Yang, Chung-Chi Chen, Hen-Hsen Huang, Hsin-Hsi Chen
cs.CL
Fake news and misinformation spread rapidly on the Internet. How to identify it and how to interpret the identification results have become important issues. In this paper, we propose a Dual Co-Attention Network (Dual-CAN) for fake news detection, which takes news content, social media replies, and external knowledge i...
http://arxiv.org/pdf/2302.03475v1
2023-02-07T14:00:40
2023-02-07T14:00:40
{ "title": [ 0.01008606, 0.02168274, -0.022369385, -0.03564453, 0.02432251, -0.009918213, 0.005027771, -0.024032593, -0.022445679, 0.04324341, -0.019226074, 0.020141602, 0.03479004, -0.004299164, -0.015258789, 0.007751465, -0.0385437, 0.051635742...
http://arxiv.org/abs/2302.04096v1
Real-Word Error Correction with Trigrams: Correcting Multiple Errors in a Sentence
Seyed MohammadSadegh Dashti
cs.CL, cs.AI
Spelling correction is a fundamental task in Text Mining. In this study, we assess the real-word error correction model proposed by Mays, Damerau and Mercer and describe several drawbacks of the model. We propose a new variation which focuses on detecting and correcting multiple real-word errors in a sentence, by manip...
http://arxiv.org/pdf/2302.04096v1
2023-02-07T13:52:14
2023-02-07T13:52:14
{ "title": [ 0.0021343231, -0.026870728, -0.02041626, -0.0262146, -0.035858154, -0.011940002, -0.0038776398, -0.008033752, -0.03326416, 0.06506348, -0.026275635, 0.005962372, 0.03527832, -0.004623413, -0.047058105, -0.017807007, -0.02067566, -0.0...
http://arxiv.org/abs/2302.03353v1
What do Language Models know about word senses? Zero-Shot WSD with Language Models and Domain Inventories
Oscar Sainz, Oier Lopez de Lacalle, Eneko Agirre, German Rigau
cs.CL
Language Models are the core for almost any Natural Language Processing system nowadays. One of their particularities is their contextualized representations, a game changer feature when a disambiguation between word senses is necessary. In this paper we aim to explore to what extent language models are capable of disc...
http://arxiv.org/pdf/2302.03353v1
2023-02-07T09:55:07
2023-02-07T09:55:07
{ "title": [ -0.02456665, -0.010597229, -0.042541504, -0.030685425, -0.05090332, -0.024047852, -0.0491333, 0.013336182, 0.018310547, 0.047302246, -0.03375244, 0.00037837029, 0.03326416, -0.02784729, -0.034332275, 0.005882263, 0.005756378, 0.05404...
http://arxiv.org/abs/2302.03341v1
The Effect of Metadata on Scientific Literature Tagging: A Cross-Field Cross-Model Study
Yu Zhang, Bowen Jin, Qi Zhu, Yu Meng, Jiawei Han
cs.DL, cs.CL, cs.LG
Due to the exponential growth of scientific publications on the Web, there is a pressing need to tag each paper with fine-grained topics so that researchers can track their interested fields of study rather than drowning in the whole literature. Scientific literature tagging is beyond a pure multi-label text classifica...
http://arxiv.org/pdf/2302.03341v1
2023-02-07T09:34:41
2023-02-07T09:34:41
{ "title": [ -0.007827759, 0.01878357, -0.049865723, -0.059509277, 0.01890564, -0.022369385, -0.020477295, -0.019363403, -0.016967773, 0.054016113, 0.0042686462, 0.0154953, 0.007030487, -0.00881958, -0.06384277, 0.007663727, -0.022766113, 0.03729...
http://arxiv.org/abs/2302.03338v1
Learning Manner of Execution from Partial Corrections
Mattias Appelgren, Alex Lascarides
cs.CL
Some actions must be executed in different ways depending on the context. For example, wiping away marker requires vigorous force while wiping away almonds requires more gentle force. In this paper we provide a model where an agent learns which manner of action execution to use in which context, drawing on evidence fro...
http://arxiv.org/pdf/2302.03338v1
2023-02-07T09:25:58
2023-02-07T09:25:58
{ "title": [ -0.022018433, -0.009033203, -0.02671814, -0.039367676, -0.013824463, -0.011711121, -0.031188965, 0.011238098, -0.026260376, 0.03878784, -0.007106781, -0.0021190643, -0.018005371, -0.028060913, -0.023330688, -0.012329102, 0.013771057, ...
http://arxiv.org/abs/2302.03297v1
AutoWS: Automated Weak Supervision Framework for Text Classification
Abhinav Bohra, Huy Nguyen, Devashish Khatwani
cs.CL
Creating large, good quality labeled data has become one of the major bottlenecks for developing machine learning applications. Multiple techniques have been developed to either decrease the dependence of labeled data (zero/few-shot learning, weak supervision) or to improve the efficiency of labeling process (active le...
http://arxiv.org/pdf/2302.03297v1
2023-02-07T07:12:05
2023-02-07T07:12:05
{ "title": [ -0.0054740906, -0.01209259, -0.019088745, -0.05505371, 0.0063285828, 0.0012216568, -0.024490356, -0.023590088, 0.002981186, 0.020965576, -0.039794922, 0.031707764, -0.00844574, 0.0016231537, -0.035614014, -0.009559631, -0.022750854, ...
http://arxiv.org/abs/2302.03269v3
PLACES: Prompting Language Models for Social Conversation Synthesis
Maximillian Chen, Alexandros Papangelis, Chenyang Tao, Seokhwan Kim, Andy Rosenbaum, Yang Liu, Zhou Yu, Dilek Hakkani-Tur
cs.CL, cs.AI, cs.IR
Collecting high quality conversational data can be very expensive for most applications and infeasible for others due to privacy, ethical, or similar concerns. A promising direction to tackle this problem is to generate synthetic dialogues by prompting large language models. In this work, we use a small set of expert-w...
http://arxiv.org/pdf/2302.03269v3
2023-02-17T04:11:53
2023-02-07T05:48:16
{ "title": [ -0.01272583, -0.0038089752, -0.017089844, -0.07739258, -0.00020730495, 0.023330688, -0.07373047, 0.0030403137, 0.009994507, 0.013557434, -0.04034424, 0.015510559, -0.002670288, -0.009101868, -0.036743164, -0.0023040771, -0.020217896, ...
http://arxiv.org/abs/2302.03241v4
Continual Pre-training of Language Models
Zixuan Ke, Yijia Shao, Haowei Lin, Tatsuya Konishi, Gyuhak Kim, Bing Liu
cs.CL, cs.AI, cs.LG, cs.NE
Language models (LMs) have been instrumental for the rapid advance of natural language processing. This paper studies continual pre-training of LMs, in particular, continual domain-adaptive pre-training (or continual DAP-training). Existing research has shown that further pre-training an LM using a domain corpus to ada...
http://arxiv.org/pdf/2302.03241v4
2023-04-12T10:36:44
2023-02-07T03:57:55
{ "title": [ 0.0042915344, 0.00048732758, -0.034576416, -0.096069336, -0.013389587, -0.008934021, -0.07940674, 0.015777588, -0.011634827, 0.04650879, -0.01461792, 0.0025196075, 0.02949524, -0.034942627, -0.015579224, 0.028625488, 0.010795593, 0.0...
http://arxiv.org/abs/2302.03222v1
Bringing the State-of-the-Art to Customers: A Neural Agent Assistant Framework for Customer Service Support
Stephen Obadinma, Faiza Khan Khattak, Shirley Wang, Tania Sidhom, Elaine Lau, Sean Robertson, Jingcheng Niu, Winnie Au, Alif Munim, Karthik Raja K. Bhaskar, Bencheng Wei, Iris Ren, Waqar Muhammad, Erin Li, Bukola Ishola, Michael Wang, Griffin Tanner, Yu-Jia Shiah, Sean X. Zhang, Kwesi P. Apponsah, Kanishk Patel, Jaswin...
cs.CL
Building Agent Assistants that can help improve customer service support requires inputs from industry users and their customers, as well as knowledge about state-of-the-art Natural Language Processing (NLP) technology. We combine expertise from academia and industry to bridge the gap and build task/domain-specific Neu...
http://arxiv.org/pdf/2302.03222v1
2023-02-07T03:11:06
2023-02-07T03:11:06
{ "title": [ -0.00011008978, -0.0039749146, -0.02407837, -0.04727173, 0.018478394, -0.011955261, -0.004348755, -0.0048942566, 0.0038261414, 0.04647827, -0.027053833, -0.009994507, 0.024810791, -0.004142761, -0.058532715, 0.0004930496, -0.004211426, ...
http://arxiv.org/abs/2302.03205v1
An entity-guided text summarization framework with relational heterogeneous graph neural network
Jingqiang Chen
cs.CL
Two crucial issues for text summarization to generate faithful summaries are to make use of knowledge beyond text and to make use of cross-sentence relations in text. Intuitive ways for the two issues are Knowledge Graph (KG) and Graph Neural Network (GNN) respectively. Entities are semantic units in text and in KG. Th...
http://arxiv.org/pdf/2302.03205v1
2023-02-07T02:27:21
2023-02-07T02:27:21
{ "title": [ -0.01449585, 0.008453369, -0.03463745, -0.066345215, -0.013793945, 0.0018024445, -0.061279297, 0.00047779083, -0.006958008, 0.058380127, 0.006904602, 0.03439331, 0.021377563, -0.0075302124, -0.038208008, 0.018249512, -0.0028572083, 0...
http://arxiv.org/abs/2302.03202v2
Exploring the Benefits of Training Expert Language Models over Instruction Tuning
Joel Jang, Seungone Kim, Seonghyeon Ye, Doyoung Kim, Lajanugen Logeswaran, Moontae Lee, Kyungjae Lee, Minjoon Seo
cs.CL
Recently, Language Models (LMs) instruction-tuned on multiple tasks, also known as multitask-prompted fine-tuning (MT), have shown the capability to generalize to unseen tasks. Previous work has shown that scaling the number of training tasks is the key component in making stronger MT LMs. In this work, we report an un...
http://arxiv.org/pdf/2302.03202v2
2023-02-09T04:55:39
2023-02-07T02:24:30
{ "title": [ -0.026779175, -0.011047363, -0.024307251, -0.062042236, -0.026382446, -0.008476257, -0.06738281, 0.0016012192, -0.03387451, 0.024780273, 0.0050239563, 0.0016469955, -0.021392822, -0.036499023, -0.015670776, -0.00497818, -0.00868988, ...
http://arxiv.org/abs/2302.03194v2
UDApter -- Efficient Domain Adaptation Using Adapters
Bhavitvya Malik, Abhinav Ramesh Kashyap, Min-Yen Kan, Soujanya Poria
cs.CL
We propose two methods to make unsupervised domain adaptation (UDA) more parameter efficient using adapters, small bottleneck layers interspersed with every layer of the large-scale pre-trained language model (PLM). The first method deconstructs UDA into a two-step process: first by adding a domain adapter to learn dom...
http://arxiv.org/pdf/2302.03194v2
2023-02-16T08:17:56
2023-02-07T02:04:17
{ "title": [ -0.022201538, -0.0044937134, -0.052490234, -0.060760498, 0.024719238, -0.019241333, -0.022735596, -0.016983032, -0.0048828125, 0.035614014, 0.017166138, -0.015365601, 0.036010742, -0.034851074, -0.018173218, 0.025970459, -0.012809753, ...
http://arxiv.org/abs/2302.03183v1
Capturing Topic Framing via Masked Language Modeling
Xiaobo Guo, Weicheng Ma, Soroush Vosoughi
cs.CL
Differential framing of issues can lead to divergent world views on important issues. This is especially true in domains where the information presented can reach a large audience, such as traditional and social media. Scalable and reliable measurement of such differential framing is an important first step in addressi...
http://arxiv.org/pdf/2302.03183v1
2023-02-07T01:23:14
2023-02-07T01:23:14
{ "title": [ -0.021911621, 0.008735657, -0.040161133, -0.04626465, -0.008834839, -0.0031490326, -0.043914795, -0.026367188, -0.0063095093, 0.02986145, 0.00064468384, 0.020126343, 0.016036987, -0.030838013, -0.013641357, 0.020477295, -0.012718201, ...
http://arxiv.org/abs/2302.06474v1
Applying BERT and ChatGPT for Sentiment Analysis of Lyme Disease in Scientific Literature
Teo Susnjak
cs.CL
This chapter presents a practical guide for conducting Sentiment Analysis using Natural Language Processing (NLP) techniques in the domain of tick-borne disease text. The aim is to demonstrate the process of how the presence of bias in the discourse surrounding chronic manifestations of the disease can be evaluated. Th...
http://arxiv.org/pdf/2302.06474v1
2023-02-07T01:15:05
2023-02-07T01:15:05
{ "title": [ 0.010292053, -0.01979065, -0.026046753, -0.03086853, -0.016204834, -0.0010251999, -0.008125305, -0.0059661865, -0.024841309, 0.027618408, -0.035888672, 0.017913818, -0.025436401, -0.0022411346, -0.02684021, -0.015060425, -0.017929077, ...
http://arxiv.org/abs/2302.03169v2
Data Selection for Language Models via Importance Resampling
Sang Michael Xie, Shibani Santurkar, Tengyu Ma, Percy Liang
cs.CL, cs.LG
Selecting a suitable pretraining dataset is crucial for both general-domain (e.g., GPT-3) and domain-specific (e.g., Codex) language models (LMs). We formalize this problem as selecting a subset of a large raw unlabeled dataset to match a desired target distribution given some unlabeled target samples. Due to the large...
http://arxiv.org/pdf/2302.03169v2
2023-10-24T17:39:05
2023-02-06T23:57:56
{ "title": [ -0.051330566, -0.008850098, -0.0692749, -0.03302002, -0.0035133362, -0.011497498, -0.056610107, 0.011734009, 0.0019664764, 0.030334473, -0.018615723, 0.012786865, 0.0049972534, -0.014389038, -0.0524292, 0.002702713, -0.0236969, 0.055...
http://arxiv.org/abs/2302.03162v3
Protecting Language Generation Models via Invisible Watermarking
Xuandong Zhao, Yu-Xiang Wang, Lei Li
cs.CR, cs.CL, cs.LG
Language generation models have been an increasingly powerful enabler for many applications. Many such models offer free or affordable API access, which makes them potentially vulnerable to model extraction attacks through distillation. To protect intellectual property (IP) and ensure fair use of these models, various ...
http://arxiv.org/pdf/2302.03162v3
2023-08-27T00:13:09
2023-02-06T23:42:03
{ "title": [ -0.030151367, 0.0019302368, -0.046325684, -0.051818848, 0.0071487427, -0.0009860992, -0.051452637, -0.0107040405, -0.025299072, 0.030319214, 0.0036182404, 0.0013885498, 0.0016317368, -0.033325195, -0.03189087, -0.0054740906, -0.030197144...
http://arxiv.org/abs/2302.03147v1
It's about Time: Rethinking Evaluation on Rumor Detection Benchmarks using Chronological Splits
Yida Mu, Kalina Bontcheva, Nikolaos Aletras
cs.CL
New events emerge over time influencing the topics of rumors in social media. Current rumor detection benchmarks use random splits as training, development and test sets which typically results in topical overlaps. Consequently, models trained on random splits may not perform well on rumor classification on previously ...
http://arxiv.org/pdf/2302.03147v1
2023-02-06T22:53:13
2023-02-06T22:53:13
{ "title": [ -0.01725769, -0.020202637, -0.022247314, -0.027999878, -0.017944336, -0.009208679, -0.0031719208, -0.017166138, 0.0065574646, 0.04547119, -0.050445557, 0.027236938, -0.029129028, 0.0061836243, -0.07556152, -0.028182983, -0.017074585, ...
http://arxiv.org/abs/2302.03145v1
Techniques to Improve Neural Math Word Problem Solvers
Youyuan Zhang
cs.CL, cs.SC
Developing automatic Math Word Problem (MWP) solvers is a challenging task that demands the ability of understanding and mathematical reasoning over the natural language. Recent neural-based approaches mainly encode the problem text using a language model and decode a mathematical expression over quantities and operato...
http://arxiv.org/pdf/2302.03145v1
2023-02-06T22:41:51
2023-02-06T22:41:51
{ "title": [ -0.0011529922, 0.007785797, -0.021240234, 0.002998352, -0.008773804, -0.036712646, 0.009292603, -0.021347046, -0.0053100586, 0.042663574, -0.02760315, 0.0031280518, 0.0209198, -0.012107849, -0.019958496, 0.009391785, 0.005355835, 0.0...
http://arxiv.org/abs/2302.03126v1
Context-Gloss Augmentation for Improving Arabic Target Sense Verification
Sanad Malaysha, Mustafa Jarrar, Mohammed Khalilia
cs.CL
Arabic language lacks semantic datasets and sense inventories. The most common semantically-labeled dataset for Arabic is the ArabGlossBERT, a relatively small dataset that consists of 167K context-gloss pairs (about 60K positive and 107K negative pairs), collected from Arabic dictionaries. This paper presents an enric...
http://arxiv.org/pdf/2302.03126v1
2023-02-06T21:24:02
2023-02-06T21:24:02
{ "title": [ 0.005924225, -0.010009766, -0.036224365, -0.07720947, 0.018936157, 0.007949829, -0.020324707, 0.01701355, -0.008720398, 0.05154419, -0.011161804, 0.019607544, 0.005935669, -0.013511658, -0.04071045, 0.009284973, 0.0009918213, -0.0023...
http://arxiv.org/abs/2302.03106v2
Efficient and Flexible Topic Modeling using Pretrained Embeddings and Bag of Sentences
Johannes Schneider
cs.CL, cs.LG
Pre-trained language models have led to a new state-of-the-art in many NLP tasks. However, for topic modeling, statistical generative models such as LDA are still prevalent, which do not easily allow incorporating contextual word vectors. They might yield topics that do not align very well with human judgment. In this ...
http://arxiv.org/pdf/2302.03106v2
2023-02-18T22:06:00
2023-02-06T20:13:11
{ "title": [ -0.00065135956, -0.00390625, -0.038116455, -0.06713867, 0.012428284, 0.0015115738, -0.039123535, -0.03010559, 0.02482605, 0.012794495, -0.003068924, 0.030410767, -0.020507812, -0.022903442, -0.018081665, 0.021621704, -0.0061798096, 0...
http://arxiv.org/abs/2302.03488v2
APAM: Adaptive Pre-training and Adaptive Meta Learning in Language Model for Noisy Labels and Long-tailed Learning
Sunyi Chi, Bo Dong, Yiming Xu, Zhenyu Shi, Zheng Du
cs.CL, cs.AI
Practical natural language processing (NLP) tasks are commonly long-tailed with noisy labels. Those problems challenge the generalization and robustness of complex models such as Deep Neural Networks (DNNs). Some commonly used resampling techniques, such as oversampling or undersampling, could easily lead to overfittin...
http://arxiv.org/pdf/2302.03488v2
2023-05-02T22:29:38
2023-02-06T18:40:04
{ "title": [ -0.016174316, 0.0127334595, -0.020935059, -0.038391113, -0.0066490173, 0.0006966591, -0.040039062, -0.028213501, 0.00178051, 0.024887085, -0.0030937195, 0.012489319, -0.0033302307, -0.043914795, -0.02444458, 0.0065994263, 0.009361267, ...
http://arxiv.org/abs/2302.02997v2
Erasure of Unaligned Attributes from Neural Representations
Shun Shao, Yftah Ziser, Shay Cohen
cs.CL
We present the Assignment-Maximization Spectral Attribute removaL (AMSAL) algorithm, which erases information from neural representations when the information to be erased is implicit rather than directly being aligned to each input example. Our algorithm works by alternating between two steps. In one, it finds an assi...
http://arxiv.org/pdf/2302.02997v2
2023-04-18T10:34:01
2023-02-06T18:32:17
{ "title": [ 0.0035152435, 0.017028809, -0.036254883, -0.014961243, 0.015556335, -0.012573242, 0.015426636, -0.033294678, 0.005706787, 0.059417725, 0.00434494, -0.036132812, 0.021484375, -0.03314209, -0.05441284, 0.028518677, -0.0035552979, 0.055...
http://arxiv.org/abs/2302.02962v1
LoFT: Enhancing Faithfulness and Diversity for Table-to-Text Generation via Logic Form Control
Yilun Zhao, Zhenting Qi, Linyong Nan, Lorenzo Jaime Yu Flores, Dragomir Radev
cs.CL
Logical Table-to-Text (LT2T) generation is tasked with generating logically faithful sentences from tables. There currently exists two challenges in the field: 1) Faithfulness: how to generate sentences that are factually correct given the table content; 2) Diversity: how to generate multiple sentences that offer diffe...
http://arxiv.org/pdf/2302.02962v1
2023-02-06T17:49:27
2023-02-06T17:49:27
{ "title": [ -0.029052734, 0.0027370453, -0.0670166, -0.064086914, 0.011474609, 0.0006275177, -0.057891846, 0.0024700165, -0.0053367615, 0.058013916, -0.020507812, 0.036743164, 0.021362305, 0.006690979, -0.053771973, 0.0064849854, -0.008255005, 0...
http://arxiv.org/abs/2302.02900v1
Controllable Lexical Simplification for English
Kim Cheng Sheang, Daniel Ferrés, Horacio Saggion
cs.CL, cs.LG
Fine-tuning Transformer-based approaches have recently shown exciting results on sentence simplification task. However, so far, no research has applied similar approaches to the Lexical Simplification (LS) task. In this paper, we present ConLS, a Controllable Lexical Simplification system fine-tuned with T5 (a Transfor...
http://arxiv.org/pdf/2302.02900v1
2023-02-06T16:09:27
2023-02-06T16:09:27
{ "title": [ -0.034423828, -0.0079956055, -0.033355713, -0.067993164, -0.0005273819, 0.0209198, -0.05609131, 0.009086609, -0.006084442, 0.06185913, -0.007419586, 0.021255493, 0.009284973, -0.0065460205, -0.04348755, 0.021499634, -0.0011711121, 0....
http://arxiv.org/abs/2302.02888v1
Findings of the TSAR-2022 Shared Task on Multilingual Lexical Simplification
Horacio Saggion, Sanja Štajner, Daniel Ferrés, Kim Cheng Sheang, Matthew Shardlow, Kai North, Marcos Zampieri
cs.CL, cs.LG
We report findings of the TSAR-2022 shared task on multilingual lexical simplification, organized as part of the Workshop on Text Simplification, Accessibility, and Readability TSAR-2022 held in conjunction with EMNLP 2022. The task called the Natural Language Processing research community to contribute with methods to...
http://arxiv.org/pdf/2302.02888v1
2023-02-06T15:53:51
2023-02-06T15:53:51
{ "title": [ -0.019348145, -0.0052452087, -0.039001465, -0.05569458, -0.0025520325, 0.0031871796, -0.067993164, -0.0041618347, -0.0020370483, 0.05718994, 0.009086609, 0.02104187, 0.041809082, -0.030151367, -0.0033988953, 0.052459717, -0.004676819, ...
http://arxiv.org/abs/2302.02852v1
Guide the Learner: Controlling Product of Experts Debiasing Method Based on Token Attribution Similarities
Ali Modarressi, Hossein Amirkhani, Mohammad Taher Pilehvar
cs.CL
Several proposals have been put forward in recent years for improving out-of-distribution (OOD) performance through mitigating dataset biases. A popular workaround is to train a robust model by re-weighting training examples based on a secondary biased model. Here, the underlying assumption is that the biased model res...
http://arxiv.org/pdf/2302.02852v1
2023-02-06T15:21:41
2023-02-06T15:21:41
{ "title": [ 0.004562378, 0.022125244, -0.036254883, -0.022857666, 0.014823914, -0.022857666, -0.021606445, -0.054595947, -0.032073975, 0.0034866333, -0.04232788, -0.000050127506, -0.007381439, -0.04135132, 0.010612488, -0.04550171, -0.0071754456, ...
http://arxiv.org/abs/2302.02813v2
Migration Reframed? A multilingual analysis on the stance shift in Europe during the Ukrainian crisis
Sergej Wildemann, Claudia Niederée, Erick Elejalde
cs.SI, cs.CL, cs.LG
The war in Ukraine seems to have positively changed the attitude toward the critical societal topic of migration in Europe -- at least towards refugees from Ukraine. We investigate whether this impression is substantiated by how the topic is reflected in online news and social media, thus linking the representation of ...
http://arxiv.org/pdf/2302.02813v2
2023-05-19T16:44:59
2023-02-06T14:36:33
{ "title": [ 0.023529053, -0.0013952255, -0.017974854, -0.015083313, -0.0042419434, -0.035491943, 0.0036811829, -0.03414917, -0.028839111, 0.04034424, -0.027053833, 0.01878357, 0.005756378, -0.017181396, -0.02229309, -0.037597656, 0.0079422, 0.01...
http://arxiv.org/abs/2302.02780v1
Coherence and Diversity through Noise: Self-Supervised Paraphrase Generation via Structure-Aware Denoising
Rishabh Gupta, Venktesh V., Mukesh Mohania, Vikram Goyal
cs.CL, cs.LG
In this paper, we propose SCANING, an unsupervised framework for paraphrasing via controlled noise injection. We focus on the novel task of paraphrasing algebraic word problems having practical applications in online pedagogy as a means to reduce plagiarism as well as ensure understanding on the part of the student ins...
http://arxiv.org/pdf/2302.02780v1
2023-02-06T13:50:57
2023-02-06T13:50:57
{ "title": [ -0.015350342, 0.00970459, -0.019927979, -0.06628418, -0.0146484375, 0.009643555, -0.07006836, -0.0030460358, -0.009216309, 0.04586792, 0.005973816, -0.0024681091, 0.020492554, -0.021194458, -0.044830322, 0.0039711, -0.017974854, 0.05...
http://arxiv.org/abs/2302.02676v8
Chain of Hindsight Aligns Language Models with Feedback
Hao Liu, Carmelo Sferrazza, Pieter Abbeel
cs.LG, cs.CL
Learning from human preferences is important for language models to match human needs and to align with human and social values. Prior works have achieved remarkable successes by learning from human feedback to understand and follow instructions. Nonetheless, these methods are either founded on hand-picked model genera...
http://arxiv.org/pdf/2302.02676v8
2023-10-18T07:11:12
2023-02-06T10:28:16
{ "title": [ 0.0028362274, -0.0027980804, -0.025527954, -0.047790527, -0.07684326, -0.0034866333, -0.09063721, -0.008476257, -0.01574707, 0.04852295, -0.02798462, -0.02279663, 0.042297363, -0.036590576, -0.0033454895, -0.0062065125, 0.020965576, ...
http://arxiv.org/abs/2302.02655v1
Evolution of grammatical forms: some quantitative approaches
Jean-Marc Luck, Anita Mehta
cond-mat.stat-mech, cs.CL
Grammatical forms are said to evolve via two main mechanisms. These are, respectively, the `descent' mechanism, where current forms can be seen to have descended (albeit with occasional modifications) from their roots in ancient languages, and the `contact' mechanism, where evolution in a given language occurs via borr...
http://arxiv.org/pdf/2302.02655v1
2023-02-06T09:50:48
2023-02-06T09:50:48
{ "title": [ -0.04284668, 0.0071754456, -0.024780273, -0.028076172, 0.010093689, -0.0011253357, -0.01737976, -0.003255844, 0.015213013, 0.02760315, -0.014053345, 0.009681702, -0.019073486, -0.014595032, -0.051086426, 0.011924744, -0.013160706, 0....
http://arxiv.org/abs/2302.02568v2
Less is More: Understanding Word-level Textual Adversarial Attack via n-gram Frequency Descend
Ning Lu, Shengcai Liu, Zhirui Zhang, Qi Wang, Haifeng Liu, Ke Tang
cs.CL, cs.AI, cs.CR, cs.LG
Word-level textual adversarial attacks have achieved striking performance in fooling natural language processing models. However, the fundamental questions of why these attacks are effective, and the intrinsic properties of the adversarial examples (AEs), are still not well understood. This work attempts to interpret t...
http://arxiv.org/pdf/2302.02568v2
2023-02-07T08:12:45
2023-02-06T05:11:27
{ "title": [ -0.0006451607, -0.008430481, -0.010887146, -0.04449463, 0.011413574, -0.004463196, -0.0062026978, -0.008125305, -0.033294678, 0.031829834, -0.0099487305, 0.03933716, 0.00055646896, -0.009208679, -0.026901245, -0.005344391, -0.016708374, ...
http://arxiv.org/abs/2302.03494v8
A Categorical Archive of ChatGPT Failures
Ali Borji
cs.CL, cs.AI, cs.LG
Large language models have been demonstrated to be valuable in different fields. ChatGPT, developed by OpenAI, has been trained using massive amounts of data and simulates human conversation by comprehending context and generating appropriate responses. It has garnered significant attention due to its ability to effect...
http://arxiv.org/pdf/2302.03494v8
2023-04-03T20:02:26
2023-02-06T04:21:59
{ "title": [ 0.028671265, -0.029724121, -0.00023829937, -0.020095825, -0.047058105, -0.012512207, 0.001540184, -0.012069702, 0.022537231, 0.035736084, -0.03074646, 0.025253296, 0.00630188, -0.03036499, 0.0038986206, 0.028625488, 0.003376007, 0.00...
http://arxiv.org/abs/2302.02500v1
TempEL: Linking Dynamically Evolving and Newly Emerging Entities
Klim Zaporojets, Lucie-Aimee Kaffee, Johannes Deleu, Thomas Demeester, Chris Develder, Isabelle Augenstein
cs.CL
In our continuously evolving world, entities change over time and new, previously non-existing or unknown, entities appear. We study how this evolutionary scenario impacts the performance on a well established entity linking (EL) task. For that study, we introduce TempEL, an entity linking dataset that consists of time...
http://arxiv.org/pdf/2302.02500v1
2023-02-05T22:34:36
2023-02-05T22:34:36
{ "title": [ 0.000015318394, -0.016983032, -0.049194336, -0.08068848, 0.02444458, -0.0049476624, -0.028259277, -0.014976501, -0.006149292, 0.019866943, -0.0017108917, 0.0061836243, 0.0017681122, -0.024505615, -0.010520935, 0.010673523, -0.0016994476,...
http://arxiv.org/abs/2302.02479v1
Hatemongers ride on echo chambers to escalate hate speech diffusion
Vasu Goel, Dhruv Sahnan, Subhabrata Dutta, Anil Bandhakavi, Tanmoy Chakraborty
cs.SI, cs.AI, cs.CL, cs.CY
Recent years have witnessed a swelling rise of hateful and abusive content over online social networks. While detection and moderation of hate speech have been the early go-to countermeasures, the solution requires a deeper exploration of the dynamics of hate generation and propagation. We analyze more than 32 million ...
http://arxiv.org/pdf/2302.02479v1
2023-02-05T20:30:48
2023-02-05T20:30:48
{ "title": [ 0.07287598, 0.015129089, -0.023086548, -0.046020508, -0.004322052, -0.03475952, 0.02935791, 0.00011318922, -0.07067871, 0.042999268, -0.010231018, 0.028305054, 0.024353027, -0.043304443, 0.019866943, -0.025512695, -0.017120361, 0.054...
http://arxiv.org/abs/2302.02463v3
Nationality Bias in Text Generation
Pranav Narayanan Venkit, Sanjana Gautam, Ruchi Panchanadikar, Ting-Hao 'Kenneth' Huang, Shomir Wilson
cs.CL, cs.AI
Little attention is placed on analyzing nationality bias in language models, especially when nationality is highly used as a factor in increasing the performance of social NLP models. This paper examines how a text generation model, GPT-2, accentuates pre-existing societal biases about country-based demonyms. We genera...
http://arxiv.org/pdf/2302.02463v3
2023-02-14T21:34:34
2023-02-05T19:15:33
{ "title": [ -0.010429382, -0.0017976761, -0.032073975, -0.03665161, -0.02708435, -0.018997192, -0.04348755, -0.023269653, -0.011329651, 0.049804688, -0.020477295, 0.006225586, 0.010353088, -0.012039185, -0.030044556, 0.003566742, -0.01966858, 0....
http://arxiv.org/abs/2302.02453v1
FineDeb: A Debiasing Framework for Language Models
Akash Saravanan, Dhruv Mullick, Habibur Rahman, Nidhi Hegde
cs.CL, cs.CY
As language models are increasingly included in human-facing machine learning tools, bias against demographic subgroups has gained attention. We propose FineDeb, a two-phase debiasing framework for language models that starts with contextual debiasing of embeddings learned by pretrained language models. The model is th...
http://arxiv.org/pdf/2302.02453v1
2023-02-05T18:35:21
2023-02-05T18:35:21
{ "title": [ -0.02923584, -0.0066184998, -0.0340271, -0.08337402, 0.010688782, 0.015579224, -0.094177246, -0.011581421, 0.011680603, 0.029647827, -0.010665894, -0.00957489, 0.025802612, -0.019134521, -0.051849365, 0.013320923, -0.019943237, 0.048...
http://arxiv.org/abs/2302.02419v1
deep learning of segment-level feature representation for speech emotion recognition in conversations
Jiachen Luo, Huy Phan, Joshua Reiss
cs.CL, cs.SD, eess.AS
Accurately detecting emotions in conversation is a necessary yet challenging task due to the complexity of emotions and dynamics in dialogues. The emotional state of a speaker can be influenced by many different factors, such as interlocutor stimulus, dialogue scene, and topic. In this work, we propose a conversational...
http://arxiv.org/pdf/2302.02419v1
2023-02-05T16:15:46
2023-02-05T16:15:46
{ "title": [ -0.0060577393, 0.00022566319, -0.0423584, -0.06414795, 0.0069770813, -0.014076233, -0.036590576, 0.017608643, 0.01436615, 0.031463623, -0.008285522, -0.011131287, 0.017669678, -0.028167725, -0.041748047, 0.009147644, -0.0035266876, 0...
http://arxiv.org/abs/2302.03499v1
Exploring Data Augmentation for Code Generation Tasks
Pinzhen Chen, Gerasimos Lampouras
cs.CL, cs.AI, cs.PL
Advances in natural language processing, such as transfer learning from pre-trained language models, have impacted how models are trained for programming language tasks too. Previous research primarily explored code pre-training and expanded it through multi-modality and multi-tasking, yet the data for downstream tasks...
http://arxiv.org/pdf/2302.03499v1
2023-02-05T14:30:32
2023-02-05T14:30:32
{ "title": [ -0.018112183, -0.021499634, -0.06781006, -0.06854248, 0.0037174225, -0.016464233, -0.034118652, 0.036346436, 0.025665283, 0.012306213, -0.044128418, 0.008842468, -0.017227173, -0.01473999, -0.024597168, 0.000895977, -0.008132935, 0.0...
http://arxiv.org/abs/2302.03498v2
MAC: A unified framework boosting low resource automatic speech recognition
Zeping Min, Qian Ge, Zhong Li, Weinan E
cs.CL, cs.SD, eess.AS
We propose a unified framework for low resource automatic speech recognition tasks named meta audio concatenation (MAC). It is easy to implement and can be carried out in extremely low resource environments. Mathematically, we give a clear description of MAC framework from the perspective of bayesian sampling. In this ...
http://arxiv.org/pdf/2302.03498v2
2023-02-15T11:09:06
2023-02-05T09:49:18
{ "title": [ -0.025650024, -0.008430481, -0.064331055, -0.093444824, 0.019729614, 0.0043907166, -0.06524658, 0.029052734, -0.0129852295, 0.042938232, -0.020248413, 0.01525116, 0.0034980774, -0.011451721, -0.042114258, 0.0068473816, -0.0051193237, ...
http://arxiv.org/abs/2302.03507v2
Meta-Learning Siamese Network for Few-Shot Text Classification
Chengcheng Han, Yuhe Wang, Yingnan Fu, Xiang Li, Minghui Qiu, Ming Gao, Aoying Zhou
cs.CL, cs.AI
Few-shot learning has been used to tackle the problem of label scarcity in text classification, of which meta-learning based methods have shown to be effective, such as the prototypical networks (PROTO). Despite the success of PROTO, there still exist three main problems: (1) ignore the randomness of the sampled suppor...
http://arxiv.org/pdf/2302.03507v2
2023-03-16T15:10:57
2023-02-05T07:41:17
{ "title": [ -0.008132935, -0.0113220215, -0.022613525, -0.023147583, -0.025604248, -0.0007157326, -0.01977539, -0.01134491, -0.0059127808, 0.050720215, -0.0093307495, 0.024841309, -0.026367188, -0.0067710876, -0.06573486, -0.014945984, 0.0011672974,...
http://arxiv.org/abs/2302.02291v3
A Semantic Approach to Negation Detection and Word Disambiguation with Natural Language Processing
Izunna Okpala, Guillermo Romera Rodriguez, Andrea Tapia, Shane Halse, Jess Kropczynski
cs.CL, cs.AI, cs.IR, I.2.7; I.5.1; I.7.1; I.7.2
This study aims to demonstrate the methods for detecting negations in a sentence by uniquely evaluating the lexical structure of the text via word-sense disambiguation. The proposed framework examines all the unique features in the various expressions within a text to resolve the contextual usage of all tokens and deci...
http://arxiv.org/pdf/2302.02291v3
2023-02-22T15:09:49
2023-02-05T03:58:45
{ "title": [ -0.005619049, -0.0050621033, -0.009536743, -0.0546875, 0.0025177002, 0.008766174, -0.03466797, -0.01878357, 0.008682251, 0.025939941, -0.05444336, -0.005065918, 0.02319336, -0.012901306, -0.03805542, -0.008377075, -0.021240234, 0.048...
http://arxiv.org/abs/2302.02275v1
Unleashing the True Potential of Sequence-to-Sequence Models for Sequence Tagging and Structure Parsing
Han He, Jinho D. Choi
cs.CL
Sequence-to-Sequence (S2S) models have achieved remarkable success on various text generation tasks. However, learning complex structures with S2S models remains challenging as external neural modules and additional lexicons are often supplemented to predict non-textual outputs. We present a systematic study of S2S mod...
http://arxiv.org/pdf/2302.02275v1
2023-02-05T01:37:26
2023-02-05T01:37:26
{ "title": [ -0.019607544, -0.0055656433, -0.020126343, -0.096069336, 0.009498596, 0.016189575, -0.066101074, 0.003993988, 0.0003976822, 0.019561768, -0.013122559, 0.0057525635, 0.0005683899, -0.025497437, -0.031585693, 0.009536743, -0.0010099411, ...
http://arxiv.org/abs/2302.02232v1
A Benchmark and Scoring Algorithm for Enriching Arabic Synonyms
Sana Ghanem, Mustafa Jarrar, Radi Jarrar, Ibrahim Bounhas
cs.CL
This paper addresses the task of extending a given synset with additional synonyms taking into account synonymy strength as a fuzzy value. Given a mono/multilingual synset and a threshold (a fuzzy value [0-1]), our goal is to extract new synonyms above this threshold from existing lexicons. We present twofold contribut...
http://arxiv.org/pdf/2302.02232v1
2023-02-04T20:30:32
2023-02-04T20:30:32
{ "title": [ -0.040374756, -0.0049438477, -0.03302002, -0.027160645, -0.009361267, 0.0021972656, -0.03186035, -0.005882263, 0.016860962, 0.039855957, -0.009231567, 0.015716553, -0.0028858185, -0.017868042, -0.05303955, 0.011070251, -0.0023231506, ...
http://arxiv.org/abs/2302.02178v1
Construction Grammar Provides Unique Insight into Neural Language Models
Leonie Weissweiler, Taiqi He, Naoki Otani, David R. Mortensen, Lori Levin, Hinrich Schütze
cs.CL
Construction Grammar (CxG) has recently been used as the basis for probing studies that have investigated the performance of large pretrained language models (PLMs) with respect to the structure and meaning of constructions. In this position paper, we make suggestions for the continuation and augmentation of this line ...
http://arxiv.org/pdf/2302.02178v1
2023-02-04T15:06:26
2023-02-04T15:06:26
{ "title": [ -0.005771637, -0.01966858, -0.04031372, -0.049835205, -0.053222656, -0.021697998, -0.05960083, 0.009262085, -0.0440979, 0.027114868, -0.021209717, -0.0024032593, -0.010002136, -0.04348755, 0.014411926, -0.018630981, -0.0067481995, 0....
http://arxiv.org/abs/2302.02169v2
How Many and Which Training Points Would Need to be Removed to Flip this Prediction?
Jinghan Yang, Sarthak Jain, Byron C. Wallace
cs.LG, cs.AI, cs.CL
We consider the problem of identifying a minimal subset of training data $\mathcal{S}_t$ such that if the instances comprising $\mathcal{S}_t$ had been removed prior to training, the categorization of a given test point $x_t$ would have been different. Identifying such a set may be of interest for a few reasons. First,...
http://arxiv.org/pdf/2302.02169v2
2023-02-09T04:23:22
2023-02-04T13:55:12
{ "title": [ 0.03286743, 0.014793396, -0.0014390945, 0.041015625, 0.0040893555, -0.03652954, 0.055786133, -0.038330078, -0.020263672, 0.027160645, 0.010604858, -0.015304565, -0.01725769, -0.036834717, -0.0051879883, -0.01838684, 0.0021362305, 0.0...
http://arxiv.org/abs/2302.02149v1
Invariants for neural automata
Jone Uria-Albizuri, Giovanni Sirio Carmantini, Peter beim Graben, Serafim Rodrigues
cs.NE, cs.CL, cs.FL, cs.SC
Computational modeling of neurodynamical systems often deploys neural networks and symbolic dynamics. A particular way for combining these approaches within a framework called vector symbolic architectures leads to neural automata. An interesting research direction we have pursued under this framework has been to consi...
http://arxiv.org/pdf/2302.02149v1
2023-02-04T11:40:40
2023-02-04T11:40:40
{ "title": [ -0.029647827, 0.02015686, -0.039123535, -0.013031006, 0.018737793, -0.0062332153, -0.008224487, -0.014076233, 0.004760742, 0.04055786, -0.014671326, -0.0018901825, -0.00856781, -0.010894775, -0.045837402, 0.012184143, -0.0007710457, ...
http://arxiv.org/abs/2302.02141v1
LipFormer: Learning to Lipread Unseen Speakers based on Visual-Landmark Transformers
Feng Xue, Yu Li, Deyin Liu, Yincen Xie, Lin Wu, Richang Hong
cs.CV, cs.CL, cs.MM
Lipreading refers to understanding and further translating the speech of a speaker in the video into natural language. State-of-the-art lipreading methods excel in interpreting overlap speakers, i.e., speakers appear in both training and inference sets. However, generalizing these methods to unseen speakers incurs cata...
http://arxiv.org/pdf/2302.02141v1
2023-02-04T10:22:18
2023-02-04T10:22:18
{ "title": [ 0.007659912, 0.011154175, -0.019821167, -0.04309082, -0.0112838745, -0.0045814514, -0.02482605, -0.023269653, 0.013374329, 0.031097412, -0.053100586, -0.0035820007, -0.0017004013, -0.0044441223, -0.058502197, -0.03314209, 0.006916046, ...
http://arxiv.org/abs/2302.02128v1
Interaction Order Prediction for Temporal Graphs
Nayana Bannur, Mashrin Srivastava, Harsha Vardhan
cs.SI, cs.CL, cs.LG
Link prediction in graphs is a task that has been widely investigated. It has been applied in various domains such as knowledge graph completion, content/item recommendation, social network recommendations and so on. The initial focus of most research was on link prediction in static graphs. However, there has recently...
http://arxiv.org/pdf/2302.02128v1
2023-02-04T08:19:13
2023-02-04T08:19:13
{ "title": [ 0.015396118, 0.001253128, -0.0690918, -0.121398926, 0.01890564, -0.011230469, -0.06958008, 0.01576233, -0.00015604496, 0.036102295, 0.009109497, 0.0033550262, -0.012229919, -0.020248413, -0.056365967, 0.001531601, -0.0078048706, 0.03...
http://arxiv.org/abs/2302.02123v3
Greedy Ordering of Layer Weight Matrices in Transformers Improves Translation
Elicia Ye
cs.CL, cs.AI, cs.NE
Prior work has attempted to understand the internal structures and functionalities of Transformer-based encoder-decoder architectures on the level of multi-head attention and feed-forward sublayers. Interpretations have focused on the encoder and decoder, along with the combinatorial possibilities of the self-attention...
http://arxiv.org/pdf/2302.02123v3
2023-03-17T00:08:53
2023-02-04T07:44:23
{ "title": [ -0.030822754, -0.00217247, -0.014961243, -0.040618896, -0.027648926, 0.010726929, -0.052093506, -0.012001038, -0.035308838, 0.057495117, -0.005832672, 0.015945435, 0.021392822, -0.02848816, -0.013900757, 0.0079574585, 0.020553589, 0....
http://arxiv.org/abs/2302.02122v1
A New cross-domain strategy based XAI models for fake news detection
Deepak Kanneganti
cs.CL, cs.AI
In this study, we presented a four-level cross-domain strategy for fake news detection on pre-trained models. Cross-domain text classification is a task of a model adopting a target domain by using the knowledge of the source domain. Explainability is crucial in understanding the behaviour of these complex models. A fi...
http://arxiv.org/pdf/2302.02122v1
2023-02-04T07:36:17
2023-02-04T07:36:17
{ "title": [ -0.019622803, 0.009597778, -0.034576416, -0.0015354156, 0.01576233, -0.0211792, -0.0073165894, -0.03237915, -0.03491211, 0.022888184, -0.015037537, 0.02796936, -0.005718231, -0.020721436, -0.013999939, -0.0046691895, -0.05316162, 0.0...
http://arxiv.org/abs/2302.02116v1
Knowledge Graph Completion Method Combined With Adaptive Enhanced Semantic Information
Weidong Ji, Zengxiang Yin, Guohui Zhou, Yuqi Yue, Xinru Zhang, Chenghong Sun
cs.CL
Translation models tend to ignore the rich semantic information in triads in the process of knowledge graph complementation. To remedy this shortcoming, this paper constructs a knowledge graph complementation method that incorporates adaptively enhanced semantic information. The hidden semantic information inherent in ...
http://arxiv.org/pdf/2302.02116v1
2023-02-04T07:01:02
2023-02-04T07:01:02
{ "title": [ -0.02178955, -0.0036449432, -0.08203125, -0.10028076, 0.01008606, 0.0013618469, -0.05657959, 0.021484375, 0.004512787, 0.03262329, -0.013977051, 0.020126343, 0.008094788, -0.013900757, -0.024749756, 0.014877319, 0.010383606, 0.043273...
http://arxiv.org/abs/2302.03496v1
Sentiment Analysis on YouTube Smart Phone Unboxing Video Reviews in Sri Lanka
Sherina Sally
cs.CL
Product-related reviews are based on users' experiences that are mostly shared on videos in YouTube. It is the second most popular website globally in 2021. People prefer to watch videos on recently released products prior to purchasing, in order to gather overall feedback and make worthy decisions. These videos are cr...
http://arxiv.org/pdf/2302.03496v1
2023-02-04T06:55:24
2023-02-04T06:55:24
{ "title": [ 0.0011262894, 0.015823364, -0.035125732, 0.0003194809, 0.0009899139, -0.021347046, 0.010375977, -0.0032253265, -0.038757324, 0.05291748, -0.026031494, 0.017990112, 0.01737976, -0.016113281, -0.023117065, -0.044525146, 0.023986816, 0....
http://arxiv.org/abs/2303.07205v3
The Science of Detecting LLM-Generated Texts
Ruixiang Tang, Yu-Neng Chuang, Xia Hu
cs.CL, cs.AI
The emergence of large language models (LLMs) has resulted in the production of LLM-generated texts that is highly sophisticated and almost indistinguishable from texts written by humans. However, this has also sparked concerns about the potential misuse of such texts, such as spreading misinformation and causing disru...
http://arxiv.org/pdf/2303.07205v3
2023-06-02T19:24:17
2023-02-04T04:49:17
{ "title": [ -0.017471313, 0.005317688, -0.037719727, 0.0056266785, -0.06640625, -0.0024604797, -0.046966553, -0.023010254, -0.05532837, 0.041656494, -0.0074310303, 0.033172607, 0.011581421, -0.036590576, 0.03277588, -0.0101623535, -0.018295288, ...
http://arxiv.org/abs/2302.02083v5
Theory of Mind Might Have Spontaneously Emerged in Large Language Models
Michal Kosinski
cs.CL, cs.CY, cs.HC
We explore the intriguing possibility that theory of mind (ToM), or the uniquely human ability to impute unobservable mental states to others, might have spontaneously emerged in large language models (LLMs). We designed 40 false-belief tasks, considered a gold standard in testing ToM in humans, and administered them t...
http://arxiv.org/pdf/2302.02083v5
2023-11-11T23:05:44
2023-02-04T03:50:01
{ "title": [ 0.0043296814, -0.032836914, -0.013725281, -0.0027503967, -0.057037354, -0.02268982, 0.010665894, 0.02558899, -0.019500732, 0.020812988, -0.022338867, -0.012145996, 0.0112838745, -0.033203125, 0.021850586, -0.016784668, -0.0072021484, ...
http://arxiv.org/abs/2302.02080v1
Improving Prediction Backward-Compatiblility in NLP Model Upgrade with Gated Fusion
Yi-An Lai, Elman Mansimov, Yuqing Xie, Yi Zhang
cs.CL
When upgrading neural models to a newer version, new errors that were not encountered in the legacy version can be introduced, known as regression errors. This inconsistent behavior during model upgrade often outweighs the benefits of accuracy gain and hinders the adoption of new models. To mitigate regression errors f...
http://arxiv.org/pdf/2302.02080v1
2023-02-04T03:40:35
2023-02-04T03:40:35
{ "title": [ -0.009231567, 0.017715454, -0.036712646, -0.06768799, -0.01348114, 0.002002716, -0.0791626, -0.026748657, 0.009796143, 0.034484863, -0.013900757, -0.0030822754, 0.013832092, -0.018173218, -0.04626465, 0.010917664, -0.02696228, 0.0816...
http://arxiv.org/abs/2302.02078v2
FGSI: Distant Supervision for Relation Extraction method based on Fine-Grained Semantic Information
Chenghong Sun, Weidong Ji, Guohui Zhou, Hui Guo, Zengxiang Yin, Yuqi Yue
cs.CL, cs.AI
The main purpose of relation extraction is to extract the semantic relationships between tagged pairs of entities in a sentence, which plays an important role in the semantic understanding of sentences and the construction of knowledge graphs. In this paper, we propose that the key semantic information within a sentenc...
http://arxiv.org/pdf/2302.02078v2
2023-03-18T08:29:52
2023-02-04T03:30:07
{ "title": [ -0.010131836, -0.01361084, -0.06112671, -0.08331299, 0.021530151, 0.008323669, -0.039916992, 0.0050849915, 0.025512695, -0.00071525574, -0.036346436, 0.0015773773, -0.00504303, -0.009346008, -0.033172607, -0.020935059, -0.012451172, ...
http://arxiv.org/abs/2302.02069v2
Heterogeneous Federated Knowledge Graph Embedding Learning and Unlearning
Xiangrong Zhu, Guangyao Li, Wei Hu
cs.LG, cs.CL
Federated Learning (FL) recently emerges as a paradigm to train a global machine learning model across distributed clients without sharing raw data. Knowledge Graph (KG) embedding represents KGs in a continuous vector space, serving as the backbone of many knowledge-driven applications. As a promising combination, fede...
http://arxiv.org/pdf/2302.02069v2
2023-02-26T04:00:43
2023-02-04T02:44:48
{ "title": [ -0.006351471, 0.03213501, -0.08905029, -0.05718994, 0.022125244, -0.027893066, -0.031463623, -0.0015325546, 0.0021762848, 0.056518555, -0.045715332, 0.032714844, 0.006893158, 0.0026073456, -0.020904541, -0.00818634, 0.008148193, 0.04...
http://arxiv.org/abs/2302.02064v2
Lived Experience Matters: Automatic Detection of Stigma on Social Media Toward People Who Use Substances
Salvatore Giorgi, Douglas Bellew, Daniel Roy Sadek Habib, Garrick Sherman, Joao Sedoc, Chase Smitterberg, Amanda Devoto, McKenzie Himelein-Wachowiak, Brenda Curtis
cs.CL
Stigma toward people who use substances (PWUS) is a leading barrier to seeking treatment.Further, those in treatment are more likely to drop out if they experience higher levels of stigmatization. While related concepts of hate speech and toxicity, including those targeted toward vulnerable populations, have been the f...
http://arxiv.org/pdf/2302.02064v2
2023-07-16T11:26:31
2023-02-04T02:26:08
{ "title": [ 0.028152466, 0.016738892, -0.027893066, -0.039886475, 0.025390625, -0.03036499, 0.005138397, -0.010009766, -0.012840271, 0.026672363, -0.05114746, 0.040100098, -0.0069503784, -0.0099105835, -0.01234436, -0.017227173, -0.014892578, 0....
http://arxiv.org/abs/2302.02060v1
Representation Deficiency in Masked Language Modeling
Yu Meng, Jitin Krishnan, Sinong Wang, Qifan Wang, Yuning Mao, Han Fang, Marjan Ghazvininejad, Jiawei Han, Luke Zettlemoyer
cs.CL, cs.LG
Masked Language Modeling (MLM) has been one of the most prominent approaches for pretraining bidirectional text encoders due to its simplicity and effectiveness. One notable concern about MLM is that the special $\texttt{[MASK]}$ symbol causes a discrepancy between pretraining data and downstream data as it is present ...
http://arxiv.org/pdf/2302.02060v1
2023-02-04T01:54:17
2023-02-04T01:54:17
{ "title": [ -0.037231445, 0.01386261, -0.05706787, -0.031280518, -0.0149383545, -0.008712769, -0.053894043, -0.0055160522, 0.0027561188, 0.03652954, -0.015838623, -0.021865845, 0.044647217, -0.039093018, -0.019577026, 0.021377563, -0.021987915, ...
http://arxiv.org/abs/2302.02029v1
Towards Few-Shot Identification of Morality Frames using In-Context Learning
Shamik Roy, Nishanth Sridhar Nakshatri, Dan Goldwasser
cs.CL
Data scarcity is a common problem in NLP, especially when the annotation pertains to nuanced socio-linguistic concepts that require specialized knowledge. As a result, few-shot identification of these concepts is desirable. Few-shot in-context learning using pre-trained Large Language Models (LLMs) has been recently ap...
http://arxiv.org/pdf/2302.02029v1
2023-02-03T23:26:59
2023-02-03T23:26:59
{ "title": [ 0.0345459, 0.0121154785, -0.034240723, -0.018096924, 0.017227173, -0.0390625, 0.032440186, -0.023223877, -0.0021705627, 0.029968262, -0.035339355, 0.0025863647, -0.015487671, -0.011680603, -0.040802002, -0.031951904, -0.023269653, 0....
http://arxiv.org/abs/2302.02023v1
TextShield: Beyond Successfully Detecting Adversarial Sentences in Text Classification
Lingfeng Shen, Ze Zhang, Haiyun Jiang, Ying Chen
cs.CL, cs.AI
Adversarial attack serves as a major challenge for neural network models in NLP, which precludes the model's deployment in safety-critical applications. A recent line of work, detection-based defense, aims to distinguish adversarial sentences from benign ones. However, {the core limitation of previous detection methods...
http://arxiv.org/pdf/2302.02023v1
2023-02-03T22:58:07
2023-02-03T22:58:07
{ "title": [ -0.004676819, 0.0049858093, -0.03161621, -0.034484863, 0.019592285, -0.0039367676, -0.014762878, -0.031066895, -0.01259613, 0.017227173, -0.012687683, 0.028427124, -0.032958984, -0.0039787292, -0.04486084, -0.01864624, -0.029022217, ...
http://arxiv.org/abs/2302.02016v1
Improving Interpretability via Explicit Word Interaction Graph Layer
Arshdeep Sekhon, Hanjie Chen, Aman Shrivastava, Zhe Wang, Yangfeng Ji, Yanjun Qi
cs.CL, cs.AI
Recent NLP literature has seen growing interest in improving model interpretability. Along this direction, we propose a trainable neural network layer that learns a global interaction graph between words and then selects more informative words using the learned word interactions. Our layer, we call WIGRAPH, can plug in...
http://arxiv.org/pdf/2302.02016v1
2023-02-03T21:56:32
2023-02-03T21:56:32
{ "title": [ -0.018173218, 0.007873535, -0.013114929, -0.05606079, 0.0024757385, 0.02607727, -0.06365967, -0.03387451, 0.0129852295, 0.020324707, -0.009506226, -0.0054779053, -0.015029907, -0.010955811, -0.043182373, 0.0046844482, -0.015312195, 0...
http://arxiv.org/abs/2302.03036v1
Witscript 2: A System for Generating Improvised Jokes Without Wordplay
Joe Toplyn
cs.CL, cs.AI, I.2.7; J.5
A previous paper presented Witscript, a system for generating conversational jokes that rely on wordplay. This paper extends that work by presenting Witscript 2, which uses a large language model to generate conversational jokes that rely on common sense instead of wordplay. Like Witscript, Witscript 2 is based on joke...
http://arxiv.org/pdf/2302.03036v1
2023-02-03T21:51:55
2023-02-03T21:51:55
{ "title": [ -0.0066223145, -0.021942139, -0.0519104, -0.00995636, -0.034454346, -0.022323608, -0.03237915, -0.0140686035, 0.021087646, 0.014457703, 0.013755798, -0.009819031, 0.008041382, -0.048339844, -0.014945984, 0.009086609, 0.01777649, 0.02...
http://arxiv.org/abs/2302.02008v1
Witscript: A System for Generating Improvised Jokes in a Conversation
Joe Toplyn
cs.CL, cs.AI, I.2.7; J.5
A chatbot is perceived as more humanlike and likeable if it includes some jokes in its output. But most existing joke generators were not designed to be integrated into chatbots. This paper presents Witscript, a novel joke generation system that can improvise original, contextually relevant jokes, such as humorous resp...
http://arxiv.org/pdf/2302.02008v1
2023-02-03T21:30:34
2023-02-03T21:30:34
{ "title": [ 0.006767273, -0.027114868, -0.05014038, -0.019165039, -0.044311523, -0.021972656, -0.027160645, -0.0025558472, 0.0231781, 0.021759033, -0.008781433, -0.0009608269, 0.018920898, -0.037139893, -0.034118652, -0.013664246, 0.010429382, 0...
http://arxiv.org/abs/2302.01984v1
PSST! Prosodic Speech Segmentation with Transformers
Nathan Roll, Calbert Graham, Simon Todd
cs.CL, cs.SD, eess.AS
Self-attention mechanisms have enabled transformers to achieve superhuman-level performance on many speech-to-text (STT) tasks, yet the challenge of automatic prosodic segmentation has remained unsolved. In this paper we finetune Whisper, a pretrained STT model, to annotate intonation unit (IU) boundaries by repurposin...
http://arxiv.org/pdf/2302.01984v1
2023-02-03T20:09:17
2023-02-03T20:09:17
{ "title": [ -0.0025863647, 0.00283432, -0.019546509, -0.06817627, -0.0008416176, 0.006198883, -0.045837402, -0.0066947937, -0.025558472, 0.04208374, -0.0064697266, 0.011146545, 0.011230469, -0.031799316, -0.029403687, 0.007949829, -0.0012369156, ...
http://arxiv.org/abs/2302.01973v3
Measuring The Impact Of Programming Language Distribution
Gabriel Orlanski, Kefan Xiao, Xavier Garcia, Jeffrey Hui, Joshua Howland, Jonathan Malmaud, Jacob Austin, Rishabh Singh, Michele Catasta
cs.LG, cs.CL, cs.PL
Current benchmarks for evaluating neural code models focus on only a small subset of programming languages, excluding many popular languages such as Go or Rust. To ameliorate this issue, we present the BabelCode framework for execution-based evaluation of any benchmark in any language. BabelCode enables new investigati...
http://arxiv.org/pdf/2302.01973v3
2023-05-24T16:20:33
2023-02-03T19:47:22
{ "title": [ -0.03265381, -0.0026512146, -0.0569458, -0.028167725, -0.024765015, -0.031555176, -0.013893127, 0.022125244, -0.024612427, 0.027786255, -0.022979736, 0.0034332275, -0.033203125, -0.036193848, -0.011871338, -0.017044067, 0.026855469, ...
http://arxiv.org/abs/2302.04818v1
Exploring the Cognitive Dynamics of Artificial Intelligence in the Post-COVID-19 and Learning 3.0 Era: A Case Study of ChatGPT
Lingfei Luan, Xi Lin, Wenbiao Li
cs.CY, cs.CL, cs.HC
The emergence of artificial intelligence has incited a paradigm shift across the spectrum of human endeavors, with ChatGPT serving as a catalyst for the transformation of various established domains, including but not limited to education, journalism, security, and ethics. In the post-pandemic era, the widespread adopt...
http://arxiv.org/pdf/2302.04818v1
2023-02-03T19:25:13
2023-02-03T19:25:13
{ "title": [ -0.016647339, -0.010604858, -0.04284668, -0.012687683, -0.00018203259, -0.027648926, 0.030532837, 0.0084991455, 0.0013093948, 0.044830322, -0.045043945, 0.035583496, 0.025146484, 0.008468628, -0.036956787, -0.01171875, 0.004737854, 0...
http://arxiv.org/abs/2302.01860v2
GLADIS: A General and Large Acronym Disambiguation Benchmark
Lihu Chen, Gaël Varoquaux, Fabian M. Suchanek
cs.CL
Acronym Disambiguation (AD) is crucial for natural language understanding on various sources, including biomedical reports, scientific papers, and search engine queries. However, existing acronym disambiguation benchmarks and tools are limited to specific domains, and the size of prior benchmarks is rather small. To ac...
http://arxiv.org/pdf/2302.01860v2
2023-03-13T21:41:39
2023-02-03T17:07:23
{ "title": [ -0.03857422, -0.015716553, -0.05255127, -0.05178833, 0.010726929, 0.0015296936, -0.04849243, 0.001748085, 0.0022411346, 0.041870117, -0.0009813309, 0.017059326, 0.02645874, -0.019805908, -0.025985718, 0.023880005, -0.0065345764, 0.03...
http://arxiv.org/abs/2302.01859v1
Generalizing to Unseen Elements: A Survey on Knowledge Extrapolation for Knowledge Graphs
Mingyang Chen, Wen Zhang, Yuxia Geng, Zezhong Xu, Jeff Z. Pan, Huajun Chen
cs.CL
Knowledge graphs (KGs) have become effective knowledge resources in diverse applications, and knowledge graph embedding (KGE) methods have attracted increasing attention in recent years. However, it's still challenging for conventional KGE methods to handle unseen entities or relations during the model test. Much effor...
http://arxiv.org/pdf/2302.01859v1
2023-02-03T17:05:59
2023-02-03T17:05:59
{ "title": [ -0.037719727, -0.016479492, -0.049560547, -0.059143066, -0.02267456, 0.0029659271, -0.059509277, 0.0021476746, 0.005092621, 0.009132385, -0.059509277, 0.01737976, -0.03274536, 0.0034866333, -0.0413208, -0.03152466, -0.006336212, 0.05...
http://arxiv.org/abs/2302.01849v1
Entity-Agnostic Representation Learning for Parameter-Efficient Knowledge Graph Embedding
Mingyang Chen, Wen Zhang, Zhen Yao, Yushan Zhu, Yang Gao, Jeff Z. Pan, Huajun Chen
cs.CL
We propose an entity-agnostic representation learning method for handling the problem of inefficient parameter storage costs brought by embedding knowledge graphs. Conventional knowledge graph embedding methods map elements in a knowledge graph, including entities and relations, into continuous vector spaces by assigni...
http://arxiv.org/pdf/2302.01849v1
2023-02-03T16:49:46
2023-02-03T16:49:46
{ "title": [ -0.018951416, -0.00092601776, -0.06665039, -0.07635498, 0.008773804, -0.0077285767, -0.044158936, 0.01361084, 0.015258789, 0.021499634, -0.013252258, 0.015342712, 0.004798889, -0.0129852295, -0.024963379, 0.007118225, 0.004890442, 0....
http://arxiv.org/abs/2302.01842v1
A Case Study for Compliance as Code with Graphs and Language Models: Public release of the Regulatory Knowledge Graph
Vladimir Ershov
cs.AI, cs.CL, cs.IR, cs.LG, I.2.1; I.2.4; J.1
The paper presents a study on using language models to automate the construction of executable Knowledge Graph (KG) for compliance. The paper focuses on Abu Dhabi Global Market regulations and taxonomy, involves manual tagging a portion of the regulations, training BERT-based models, which are then applied to the rest ...
http://arxiv.org/pdf/2302.01842v1
2023-02-03T16:37:08
2023-02-03T16:37:08
{ "title": [ -0.033355713, -0.026550293, -0.047027588, -0.07928467, -0.0340271, 0.012306213, -0.08648682, 0.042053223, -0.026123047, 0.016738892, -0.03878784, 0.0042266846, -0.0004963875, -0.032165527, -0.0010757446, -0.0309906, 0.0074005127, 0.0...
http://arxiv.org/abs/2302.01839v1
Investigating Stylistic Profiles for the Task of Empathy Classification in Medical Narrative Essays
Priyanka Dey, Roxana Girju
cs.CL
One important aspect of language is how speakers generate utterances and texts to convey their intended meanings. In this paper, we bring various aspects of the Construction Grammar (CxG) and the Systemic Functional Grammar (SFG) theories in a deep learning computational framework to model empathic language. Our corpus...
http://arxiv.org/pdf/2302.01839v1
2023-02-03T16:30:09
2023-02-03T16:30:09
{ "title": [ -0.039093018, -0.008796692, -0.031143188, 0.0031738281, -0.011268616, -0.009101868, 0.03010559, -0.011634827, -0.021957397, 0.007293701, 0.020065308, 0.011222839, -0.022781372, -0.03274536, -0.017349243, -0.009864807, -0.032226562, 0...
http://arxiv.org/abs/2302.01823v1
Lexical Simplification using multi level and modular approach
Nikita Katyal, Pawan Kumar Rajpoot
cs.CL
Text Simplification is an ongoing problem in Natural Language Processing, solution to which has varied implications. In conjunction with the TSAR-2022 Workshop @EMNLP2022 Lexical Simplification is the process of reducing the lexical complexity of a text by replacing difficult words with easier to read (or understand) e...
http://arxiv.org/pdf/2302.01823v1
2023-02-03T15:57:54
2023-02-03T15:57:54
{ "title": [ -0.03189087, -0.006248474, -0.06689453, -0.091308594, 0.01626587, 0.0066719055, -0.06958008, 0.03074646, -0.011474609, 0.056762695, 0.016525269, 0.013145447, 0.027145386, -0.033996582, -0.027664185, 0.028411865, -0.004737854, 0.03250...
http://arxiv.org/abs/2302.01806v1
Mitigating Data Scarcity for Large Language Models
Hoang Van
cs.CL, cs.AI
In recent years, pretrained neural language models (PNLMs) have taken the field of natural language processing by storm, achieving new benchmarks and state-of-the-art performances. These models often rely heavily on annotated data, which may not always be available. Data scarcity are commonly found in specialized domai...
http://arxiv.org/pdf/2302.01806v1
2023-02-03T15:17:53
2023-02-03T15:17:53
{ "title": [ -0.028030396, -0.0055618286, -0.050842285, -0.08380127, 0.0057411194, 0.011207581, -0.07067871, 0.009185791, 0.009353638, 0.036254883, -0.028305054, -0.0048294067, -0.001914978, -0.002128601, -0.053710938, 0.0056762695, -0.008811951, ...
http://arxiv.org/abs/2302.02801v1
LaMPP: Language Models as Probabilistic Priors for Perception and Action
Belinda Z. Li, William Chen, Pratyusha Sharma, Jacob Andreas
cs.LG, cs.CL
Language models trained on large text corpora encode rich distributional information about real-world environments and action sequences. This information plays a crucial role in current approaches to language processing tasks like question answering and instruction generation. We describe how to leverage language model...
http://arxiv.org/pdf/2302.02801v1
2023-02-03T15:14:04
2023-02-03T15:14:04
{ "title": [ -0.031829834, -0.0030536652, -0.051635742, -0.056884766, 0.00007534027, -0.009094238, -0.043823242, -0.0013542175, -0.0039863586, 0.010528564, -0.012550354, 0.0024642944, 0.0032806396, -0.022720337, -0.026672363, 0.009963989, -0.00621414...
http://arxiv.org/abs/2302.01935v2
CAB: Empathetic Dialogue Generation with Cognition, Affection and Behavior
Pan Gao, Donghong Han, Rui Zhou, Xuejiao Zhang, Zikun Wang
cs.CL, cs.HC
Empathy is an important characteristic to be considered when building a more intelligent and humanized dialogue agent. However, existing methods did not fully comprehend empathy as a complex process involving three aspects: cognition, affection and behavior. In this paper, we propose CAB, a novel framework that takes a...
http://arxiv.org/pdf/2302.01935v2
2023-02-22T00:45:26
2023-02-03T14:31:17
{ "title": [ -0.022445679, -0.021774292, -0.04360962, -0.05697632, -0.009811401, -0.0060195923, -0.029525757, 0.02192688, 0.009002686, 0.016571045, -0.0063819885, -0.012504578, 0.020355225, -0.03753662, -0.033233643, -0.00032234192, -0.01953125, ...
http://arxiv.org/abs/2302.01691v2
LIQUID: A Framework for List Question Answering Dataset Generation
Seongyun Lee, Hyunjae Kim, Jaewoo Kang
cs.CL
Question answering (QA) models often rely on large-scale training datasets, which necessitates the development of a data generation framework to reduce the cost of manual annotations. Although several recent studies have aimed to generate synthetic questions with single-span answers, no study has been conducted on the ...
http://arxiv.org/pdf/2302.01691v2
2023-02-06T08:04:56
2023-02-03T12:42:45
{ "title": [ -0.012420654, -0.0079956055, -0.049987793, -0.056488037, 0.008300781, 0.00038337708, -0.031585693, 0.009971619, 0.023803711, 0.009239197, -0.030517578, 0.005405426, -0.028259277, 0.0009737015, -0.05911255, -0.006286621, -0.0022792816, ...
http://arxiv.org/abs/2302.01676v2
Show me your NFT and I tell you how it will perform: Multimodal representation learning for NFT selling price prediction
Davide Costa, Lucio La Cava, Andrea Tagarelli
cs.LG, cs.AI, cs.CL, cs.CV, cs.NE
Non-Fungible Tokens (NFTs) represent deeds of ownership, based on blockchain technologies and smart contracts, of unique crypto assets on digital art forms (e.g., artworks or collectibles). In the spotlight after skyrocketing in 2021, NFTs have attracted the attention of crypto enthusiasts and investors intent on placi...
http://arxiv.org/pdf/2302.01676v2
2023-02-06T10:44:21
2023-02-03T11:56:38
{ "title": [ 0.014602661, 0.022247314, -0.035949707, -0.024429321, 0.009468079, -0.03164673, 0.0059776306, -0.026809692, -0.017120361, 0.02848816, -0.0049743652, -0.006137848, 0.032104492, -0.031280518, -0.01651001, -0.010368347, 0.0017671585, 0....