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/2305.08844v2
RL4F: Generating Natural Language Feedback with Reinforcement Learning for Repairing Model Outputs
Afra Feyza Akyürek, Ekin Akyürek, Aman Madaan, Ashwin Kalyan, Peter Clark, Derry Wijaya, Niket Tandon
cs.CL
Despite their unprecedented success, even the largest language models make mistakes. Similar to how humans learn and improve using feedback, previous work proposed providing language models with natural language feedback to guide them in repairing their outputs. Because human-generated critiques are expensive to obtain...
http://arxiv.org/pdf/2305.08844v2
2023-07-11T18:29:12
2023-05-15T17:57:16
{ "title": [ -0.021484375, -0.024002075, -0.018585205, -0.056518555, -0.019638062, 0.013946533, -0.05279541, -0.01486969, -0.016784668, 0.020339966, -0.016647339, 0.0043411255, -0.011703491, -0.014411926, -0.033355713, -0.0076828003, -0.01852417, ...
http://arxiv.org/abs/2305.08828v2
PMIndiaSum: Multilingual and Cross-lingual Headline Summarization for Languages in India
Ashok Urlana, Pinzhen Chen, Zheng Zhao, Shay B. Cohen, Manish Shrivastava, Barry Haddow
cs.CL, I.2.7
This paper introduces PMIndiaSum, a multilingual and massively parallel summarization corpus focused on languages in India. Our corpus provides a training and testing ground for four language families, 14 languages, and the largest to date with 196 language pairs. We detail our construction workflow including data acqu...
http://arxiv.org/pdf/2305.08828v2
2023-10-20T00:42:27
2023-05-15T17:41:15
{ "title": [ -0.03262329, -0.019592285, -0.030639648, -0.054718018, -0.026535034, 0.00579834, -0.08215332, -0.0077781677, -0.023773193, 0.046051025, -0.001657486, 0.032470703, 0.020401001, -0.003929138, -0.020019531, 0.022659302, -0.0070610046, 0...
http://arxiv.org/abs/2305.08818v1
Sentence Level Curriculum Learning for Improved Neural Conversational Models
Sean Paulsen
cs.CL, cs.LG
Designing machine intelligence to converse with a human user necessarily requires an understanding of how humans participate in conversation, and thus conversation modeling is an important task in natural language processing. New breakthroughs in architecture and data gathering continue to push the performance of such ...
http://arxiv.org/pdf/2305.08818v1
2023-05-15T17:28:59
2023-05-15T17:28:59
{ "title": [ -0.0067253113, 0.0027923584, -0.03225708, -0.044952393, -0.007820129, -0.011703491, -0.049987793, 0.00018966198, -0.0025806427, 0.019424438, -0.001865387, 0.007095337, -0.0135650635, -0.023605347, -0.022155762, 0.0028095245, -0.021987915...
http://arxiv.org/abs/2305.08809v1
Interpretability at Scale: Identifying Causal Mechanisms in Alpaca
Zhengxuan Wu, Atticus Geiger, Christopher Potts, Noah D. Goodman
cs.CL
Obtaining human-interpretable explanations of large, general-purpose language models is an urgent goal for AI safety. However, it is just as important that our interpretability methods are faithful to the causal dynamics underlying model behavior and able to robustly generalize to unseen inputs. Distributed Alignment S...
http://arxiv.org/pdf/2305.08809v1
2023-05-15T17:15:40
2023-05-15T17:15:40
{ "title": [ -0.04623413, -0.007659912, -0.04324341, -0.0435791, -0.0234375, 0.0016174316, -0.072387695, -0.011817932, -0.031051636, -0.01586914, -0.0038719177, -0.0057754517, -0.012794495, -0.05368042, 0.025665283, -0.026657104, 0.005710602, 0.0...
http://arxiv.org/abs/2305.08804v1
Exploring In-Context Learning Capabilities of Foundation Models for Generating Knowledge Graphs from Text
Hanieh Khorashadizadeh, Nandana Mihindukulasooriya, Sanju Tiwari, Jinghua Groppe, Sven Groppe
cs.CL
Knowledge graphs can represent information about the real-world using entities and their relations in a structured and semantically rich manner and they enable a variety of downstream applications such as question-answering, recommendation systems, semantic search, and advanced analytics. However, at the moment, buildi...
http://arxiv.org/pdf/2305.08804v1
2023-05-15T17:10:19
2023-05-15T17:10:19
{ "title": [ -0.037017822, -0.018203735, -0.06866455, -0.063964844, -0.015617371, 0.00040364265, -0.057769775, 0.020904541, 0.0077705383, 0.033599854, -0.021347046, 0.021408081, -0.021469116, -0.0034751892, -0.051239014, -0.014183044, -0.016967773, ...
http://arxiv.org/abs/2305.08800v1
Measuring Cross-Lingual Transferability of Multilingual Transformers on Sentence Classification
Zewen Chi, Heyan Huang, Xian-Ling Mao
cs.CL
Recent studies have exhibited remarkable capabilities of pre-trained multilingual Transformers, especially cross-lingual transferability. However, current methods do not measure cross-lingual transferability well, hindering the understanding of multilingual Transformers. In this paper, we propose IGap, a cross-lingual ...
http://arxiv.org/pdf/2305.08800v1
2023-05-15T17:05:45
2023-05-15T17:05:45
{ "title": [ -0.011306763, 0.022018433, -0.011528015, -0.03451538, 0.014045715, 0.0048103333, -0.030288696, -0.036010742, 0.00038194656, 0.027160645, -0.015792847, 0.023666382, 0.0057029724, -0.01008606, -0.03729248, 0.00655365, -0.020568848, 0.0...
http://arxiv.org/abs/2305.08787v1
Comparing Variation in Tokenizer Outputs Using a Series of Problematic and Challenging Biomedical Sentences
Christopher Meaney, Therese A Stukel, Peter C Austin, Michael Escobar
cs.CL
Background & Objective: Biomedical text data are increasingly available for research. Tokenization is an initial step in many biomedical text mining pipelines. Tokenization is the process of parsing an input biomedical sentence (represented as a digital character sequence) into a discrete set of word/token symbols, whi...
http://arxiv.org/pdf/2305.08787v1
2023-05-15T16:46:47
2023-05-15T16:46:47
{ "title": [ -0.0019426346, 0.0042800903, -0.026657104, -0.018417358, -0.016906738, -0.015304565, -0.052886963, -0.030197144, -0.018722534, 0.014762878, -0.024276733, 0.022506714, -0.035003662, -0.023010254, -0.02192688, -0.033111572, -0.032989502, ...
http://arxiv.org/abs/2305.10447v1
The Effectiveness of a Dynamic Loss Function in Neural Network Based Automated Essay Scoring
Oscar Morris
cs.CL, cs.AI
Neural networks and in particular the attention mechanism have brought significant advances to the field of Automated Essay Scoring. Many of these systems use a regression-based model which may be prone to underfitting when the model only predicts the mean of the training data. In this paper, we present a dynamic loss ...
http://arxiv.org/pdf/2305.10447v1
2023-05-15T16:39:35
2023-05-15T16:39:35
{ "title": [ -0.008132935, 0.0048217773, -0.0034255981, 0.013305664, -0.010475159, -0.011672974, 0.041015625, -0.04852295, 0.016448975, 0.020385742, -0.027038574, -0.0149002075, -0.026351929, -0.002538681, -0.070617676, -0.028656006, -0.0038909912, ...
http://arxiv.org/abs/2305.08777v1
Question-Answering System Extracts Information on Injection Drug Use from Clinical Progress Notes
Maria Mahbub, Ian Goethert, Ioana Danciu, Kathryn Knight, Sudarshan Srinivasan, Suzanne Tamang, Karine Rozenberg-Ben-Dror, Hugo Solares, Susana Martins, Edmon Begoli, Gregory D. Peterson
cs.AI, cs.CL, cs.LG
Injection drug use (IDU) is a dangerous health behavior that increases mortality and morbidity. Identifying IDU early and initiating harm reduction interventions can benefit individuals at risk. However, extracting IDU behaviors from patients' electronic health records (EHR) is difficult because there is no Internation...
http://arxiv.org/pdf/2305.08777v1
2023-05-15T16:37:00
2023-05-15T16:37:00
{ "title": [ 0.029464722, -0.008056641, -0.040252686, -0.042510986, -0.033111572, -0.016921997, -0.014755249, 0.016937256, -0.03765869, 0.0155181885, -0.06781006, 0.0027751923, -0.0039520264, -0.016998291, 0.02330017, -0.051757812, 0.018035889, 0...
http://arxiv.org/abs/2305.08732v3
Knowledge Rumination for Pre-trained Language Models
Yunzhi Yao, Peng Wang, Shengyu Mao, Chuanqi Tan, Fei Huang, Huajun Chen, Ningyu Zhang
cs.CL, cs.AI, cs.IR, cs.LG
Previous studies have revealed that vanilla pre-trained language models (PLMs) lack the capacity to handle knowledge-intensive NLP tasks alone; thus, several works have attempted to integrate external knowledge into PLMs. However, despite the promising outcome, we empirically observe that PLMs may have already encoded ...
http://arxiv.org/pdf/2305.08732v3
2023-10-11T10:51:12
2023-05-15T15:47:09
{ "title": [ -0.03869629, -0.025756836, -0.016052246, -0.05557251, -0.049224854, 0.014526367, -0.07110596, 0.01789856, 0.009284973, 0.03717041, -0.017074585, -0.019546509, 0.012550354, -0.02192688, -0.05239868, 0.0131073, 0.015823364, 0.022613525...
http://arxiv.org/abs/2305.08714v2
Sensitivity and Robustness of Large Language Models to Prompt Template in Japanese Text Classification Tasks
Chengguang Gan, Tatsunori Mori
cs.CL, cs.AI
Prompt engineering relevance research has seen a notable surge in recent years, primarily driven by advancements in pre-trained language models and large language models. However, a critical issue has been identified within this domain: the inadequate of sensitivity and robustness of these models towards Prompt Templat...
http://arxiv.org/pdf/2305.08714v2
2023-06-08T02:14:20
2023-05-15T15:19:08
{ "title": [ -0.024093628, 0.018585205, -0.023254395, -0.028015137, -0.0076942444, 0.00040864944, -0.06719971, -0.039215088, 0.015563965, 0.032440186, -0.030151367, 0.004207611, 0.012062073, -0.005748749, -0.070617676, -0.01448822, -0.007873535, ...
http://arxiv.org/abs/2305.08711v3
sustain.AI: a Recommender System to analyze Sustainability Reports
Lars Hillebrand, Maren Pielka, David Leonhard, Tobias Deußer, Tim Dilmaghani, Bernd Kliem, Rüdiger Loitz, Milad Morad, Christian Temath, Thiago Bell, Robin Stenzel, Rafet Sifa
cs.CL, cs.AI, cs.LG, H.3.3
We present sustainAI, an intelligent, context-aware recommender system that assists auditors and financial investors as well as the general public to efficiently analyze companies' sustainability reports. The tool leverages an end-to-end trainable architecture that couples a BERT-based encoding module with a multi-labe...
http://arxiv.org/pdf/2305.08711v3
2023-05-26T07:49:33
2023-05-15T15:16:19
{ "title": [ -0.013801575, 0.0015459061, -0.051574707, -0.05682373, 0.009742737, -0.015686035, -0.011299133, -0.0104522705, -0.0010299683, 0.006389618, -0.04763794, -0.006298065, -0.018493652, -0.011993408, -0.040222168, -0.028747559, -0.020812988, ...
http://arxiv.org/abs/2305.08709v1
Back Translation for Speech-to-text Translation Without Transcripts
Qingkai Fang, Yang Feng
cs.CL, cs.SD, eess.AS, I.2.7
The success of end-to-end speech-to-text translation (ST) is often achieved by utilizing source transcripts, e.g., by pre-training with automatic speech recognition (ASR) and machine translation (MT) tasks, or by introducing additional ASR and MT data. Unfortunately, transcripts are only sometimes available since numer...
http://arxiv.org/pdf/2305.08709v1
2023-05-15T15:12:40
2023-05-15T15:12:40
{ "title": [ -0.013801575, -0.016616821, -0.0038700104, -0.03918457, -0.06304932, 0.0060691833, -0.06750488, -0.0044021606, 0.0062065125, 0.06750488, 0.0030021667, -0.01574707, 0.051086426, -0.041809082, -0.013626099, 0.03036499, 0.025390625, 0.0...
http://arxiv.org/abs/2305.08706v1
Understanding and Bridging the Modality Gap for Speech Translation
Qingkai Fang, Yang Feng
cs.CL, cs.SD, eess.AS, I.2.7
How to achieve better end-to-end speech translation (ST) by leveraging (text) machine translation (MT) data? Among various existing techniques, multi-task learning is one of the effective ways to share knowledge between ST and MT in which additional MT data can help to learn source-to-target mapping. However, due to th...
http://arxiv.org/pdf/2305.08706v1
2023-05-15T15:09:18
2023-05-15T15:09:18
{ "title": [ -0.039733887, -0.0056610107, -0.02923584, -0.055603027, -0.009292603, 0.0054893494, -0.054626465, 0.0054130554, -0.01322937, 0.055389404, -0.005470276, 0.0013046265, 0.04449463, -0.03366089, -0.036895752, 0.0211792, -0.0009293556, 0....
http://arxiv.org/abs/2305.08703v4
Schema-adaptable Knowledge Graph Construction
Hongbin Ye, Honghao Gui, Xin Xu, Xi Chen, Huajun Chen, Ningyu Zhang
cs.CL, cs.AI, cs.DB, cs.IR, cs.LG
Conventional Knowledge Graph Construction (KGC) approaches typically follow the static information extraction paradigm with a closed set of pre-defined schema. As a result, such approaches fall short when applied to dynamic scenarios or domains, whereas a new type of knowledge emerges. This necessitates a system that c...
http://arxiv.org/pdf/2305.08703v4
2023-11-15T12:55:56
2023-05-15T15:06:20
{ "title": [ -0.047424316, -0.020874023, -0.07745361, -0.084106445, -0.0044441223, 0.009880066, -0.062438965, 0.031158447, -0.00121212, 0.010810852, -0.014770508, -0.012207031, -0.028900146, -0.016815186, -0.036102295, -0.01007843, 0.014556885, 0...
http://arxiv.org/abs/2305.08702v1
Recyclable Tuning for Continual Pre-training
Yujia Qin, Cheng Qian, Xu Han, Yankai Lin, Huadong Wang, Ruobing Xie, Zhiyuan Liu, Maosong Sun, Jie Zhou
cs.CL, cs.AI
Continual pre-training is the paradigm where pre-trained language models (PLMs) continually acquire fresh knowledge from growing data and gradually get upgraded. Before an upgraded PLM is released, we may have tuned the original PLM for various tasks and stored the adapted weights. However, when tuning the upgraded PLM...
http://arxiv.org/pdf/2305.08702v1
2023-05-15T15:05:44
2023-05-15T15:05:44
{ "title": [ 0.022338867, 0.0022411346, -0.040252686, -0.121520996, -0.010040283, 0.004787445, -0.08337402, 0.02684021, -0.024368286, 0.044189453, -0.015930176, -0.008094788, -0.0019893646, -0.030426025, -0.015060425, -0.0027656555, 0.029632568, ...
http://arxiv.org/abs/2305.08698v2
Continual Multimodal Knowledge Graph Construction
Xiang Chen, Ningyu Zhang, Jintian Zhang, Xiaohan Wang, Tongtong Wu, Xi Chen, Yongheng Wang, Huajun Chen
cs.CL, cs.AI, cs.DB, cs.LG, cs.MM
Multimodal Knowledge Graph Construction (MKGC) involves creating structured representations of entities and relations using multiple modalities, such as text and images. However, existing MKGC models face challenges in handling the addition of new entities and relations in dynamic real-world scenarios. The current cont...
http://arxiv.org/pdf/2305.08698v2
2023-08-01T10:23:20
2023-05-15T14:58:28
{ "title": [ -0.038238525, -0.007434845, -0.07836914, -0.09942627, -0.007259369, 0.0024414062, -0.09313965, 0.024673462, 0.0019264221, 0.033966064, 0.0071983337, 0.011688232, -0.015022278, -0.019470215, -0.037261963, 0.008728027, 0.0057907104, 0....
http://arxiv.org/abs/2305.08677v1
Natural Language Decomposition and Interpretation of Complex Utterances
Harsh Jhamtani, Hao Fang, Patrick Xia, Eran Levy, Jacob Andreas, Ben Van Durme
cs.CL
Natural language interfaces often require supervised data to translate user requests into programs, database queries, or other structured intent representations. During data collection, it can be difficult to anticipate and formalize the full range of user needs -- for example, in a system designed to handle simple req...
http://arxiv.org/pdf/2305.08677v1
2023-05-15T14:35:00
2023-05-15T14:35:00
{ "title": [ -0.0151901245, -0.0028820038, -0.02720642, -0.08862305, 0.013442993, 0.013694763, -0.0574646, 0.009757996, -0.015701294, 0.030532837, -0.008636475, 0.013725281, -0.0031433105, -0.012107849, -0.041503906, 0.0065345764, -0.005176544, 0...
http://arxiv.org/abs/2305.08655v1
Unsupervised Sentence Representation Learning with Frequency-induced Adversarial Tuning and Incomplete Sentence Filtering
Bing Wang, Ximing Li, Zhiyao Yang, Yuanyuan Guan, Jiayin Li, Shengsheng Wang
cs.CL
Pre-trained Language Model (PLM) is nowadays the mainstay of Unsupervised Sentence Representation Learning (USRL). However, PLMs are sensitive to the frequency information of words from their pre-training corpora, resulting in anisotropic embedding space, where the embeddings of high-frequency words are clustered but t...
http://arxiv.org/pdf/2305.08655v1
2023-05-15T13:59:23
2023-05-15T13:59:23
{ "title": [ -0.0051345825, 0.014404297, -0.009376526, -0.050231934, 0.0013580322, 0.012496948, -0.03515625, -0.021743774, -0.012763977, 0.036132812, -0.022399902, 0.03125, 0.009361267, -0.0154953, -0.018417358, 0.0037155151, -0.00541687, 0.04141...
http://arxiv.org/abs/2305.08654v1
Unsupervised Semantic Variation Prediction using the Distribution of Sibling Embeddings
Taichi Aida, Danushka Bollegala
cs.CL
Languages are dynamic entities, where the meanings associated with words constantly change with time. Detecting the semantic variation of words is an important task for various NLP applications that must make time-sensitive predictions. Existing work on semantic variation prediction have predominantly focused on compar...
http://arxiv.org/pdf/2305.08654v1
2023-05-15T13:58:21
2023-05-15T13:58:21
{ "title": [ -0.036834717, -0.0011920929, -0.078430176, -0.062927246, 0.011192322, -0.0024414062, -0.06732178, 0.01637268, 0.018325806, 0.02772522, -0.007633209, 0.016113281, 0.011199951, -0.020477295, -0.03173828, 0.019592285, 0.010604858, 0.069...
http://arxiv.org/abs/2305.08636v1
AdamR at SemEval-2023 Task 10: Solving the Class Imbalance Problem in Sexism Detection with Ensemble Learning
Adam Rydelek, Daryna Dementieva, Georg Groh
cs.CL, cs.AI
The Explainable Detection of Online Sexism task presents the problem of explainable sexism detection through fine-grained categorisation of sexist cases with three subtasks. Our team experimented with different ways to combat class imbalance throughout the tasks using data augmentation and loss alteration techniques. W...
http://arxiv.org/pdf/2305.08636v1
2023-05-15T13:28:59
2023-05-15T13:28:59
{ "title": [ 0.0020160675, 0.010322571, -0.059814453, -0.019424438, 0.023162842, -0.034423828, -0.014343262, -0.0211792, 0.010696411, 0.04284668, -0.025650024, 0.015022278, -0.0047340393, -0.0037059784, -0.050048828, -0.006122589, -0.036468506, 0...
http://arxiv.org/abs/2305.08633v1
Text2Gender: A Deep Learning Architecture for Analysis of Blogger's Age and Gender
Vishesh Thakur, Aneesh Tickoo
cs.CL, cs.AI
Deep learning techniques have gained a lot of traction in the field of NLP research. The aim of this paper is to predict the age and gender of an individual by inspecting their written text. We propose a supervised BERT-based classification technique in order to predict the age and gender of bloggers. The dataset used ...
http://arxiv.org/pdf/2305.08633v1
2023-05-15T13:26:50
2023-05-15T13:26:50
{ "title": [ -0.008842468, 0.003320694, -0.042633057, -0.048797607, 0.020858765, -0.02142334, -0.0041275024, 0.00063848495, 0.010414124, 0.030258179, -0.04562378, 0.019119263, 0.018112183, 0.008331299, -0.041046143, -0.0037441254, -0.005706787, 0...
http://arxiv.org/abs/2305.08625v1
Adam-Smith at SemEval-2023 Task 4: Discovering Human Values in Arguments with Ensembles of Transformer-based Models
Daniel Schroter, Daryna Dementieva, Georg Groh
cs.CL, cs.AI
This paper presents the best-performing approach alias "Adam Smith" for the SemEval-2023 Task 4: "Identification of Human Values behind Arguments". The goal of the task was to create systems that automatically identify the values within textual arguments. We train transformer-based models until they reach their loss mi...
http://arxiv.org/pdf/2305.08625v1
2023-05-15T13:20:14
2023-05-15T13:20:14
{ "title": [ 0.030532837, 0.022399902, -0.047973633, -0.007598877, 0.012748718, -0.039886475, 0.00051259995, -0.023513794, -0.025344849, 0.03729248, 0.004585266, 0.008705139, -0.006668091, -0.03491211, -0.008583069, -0.0045814514, -0.026519775, 0...
http://arxiv.org/abs/2305.08596v2
DarkBERT: A Language Model for the Dark Side of the Internet
Youngjin Jin, Eugene Jang, Jian Cui, Jin-Woo Chung, Yongjae Lee, Seungwon Shin
cs.CL, I.2.7
Recent research has suggested that there are clear differences in the language used in the Dark Web compared to that of the Surface Web. As studies on the Dark Web commonly require textual analysis of the domain, language models specific to the Dark Web may provide valuable insights to researchers. In this work, we int...
http://arxiv.org/pdf/2305.08596v2
2023-05-18T05:02:29
2023-05-15T12:23:10
{ "title": [ -0.016555786, -0.0021858215, -0.03768921, -0.039001465, 0.02154541, -0.011642456, -0.016433716, -0.00920105, 0.00073957443, 0.03704834, -0.0034122467, 0.00083875656, 0.03387451, -0.020706177, -0.042022705, 0.024978638, -0.018463135, ...
http://arxiv.org/abs/2305.10446v1
Emotion Recognition based on Psychological Components in Guided Narratives for Emotion Regulation
Gustave Cortal, Alain Finkel, Patrick Paroubek, Lina Ye
cs.CL, cs.AI
Emotion regulation is a crucial element in dealing with emotional events and has positive effects on mental health. This paper aims to provide a more comprehensive understanding of emotional events by introducing a new French corpus of emotional narratives collected using a questionnaire for emotion regulation. We foll...
http://arxiv.org/pdf/2305.10446v1
2023-05-15T12:06:31
2023-05-15T12:06:31
{ "title": [ -0.03567505, -0.0058937073, -0.03616333, -0.03186035, -0.015823364, -0.012573242, -0.016189575, 0.009742737, 0.008125305, 0.015266418, -0.016830444, -0.005973816, 0.0128479, -0.050323486, -0.023590088, -0.005847931, -0.010597229, 0.0...
http://arxiv.org/abs/2305.08572v1
Estimating the Causal Effects of Natural Logic Features in Neural NLI Models
Julia Rozanova, Marco Valentino, Andre Freitas
cs.CL
Rigorous evaluation of the causal effects of semantic features on language model predictions can be hard to achieve for natural language reasoning problems. However, this is such a desirable form of analysis from both an interpretability and model evaluation perspective, that it is valuable to zone in on specific p...
http://arxiv.org/pdf/2305.08572v1
2023-05-15T12:01:09
2023-05-15T12:01:09
{ "title": [ -0.035888672, -0.011680603, -0.04272461, -0.046844482, -0.007472992, -0.0076904297, -0.04626465, 0.0035266876, -0.013458252, 0.012870789, 0.003982544, -0.019210815, -0.0011692047, -0.04019165, -0.029373169, -0.0009551048, -0.020828247, ...
http://arxiv.org/abs/2305.08566v4
NLG Evaluation Metrics Beyond Correlation Analysis: An Empirical Metric Preference Checklist
Iftitahu Ni'mah, Meng Fang, Vlado Menkovski, Mykola Pechenizkiy
cs.CL
In this study, we analyze automatic evaluation metrics for Natural Language Generation (NLG), specifically task-agnostic metrics and human-aligned metrics. Task-agnostic metrics, such as Perplexity, BLEU, BERTScore, are cost-effective and highly adaptable to diverse NLG tasks, yet they have a weak correlation with huma...
http://arxiv.org/pdf/2305.08566v4
2023-05-26T07:30:35
2023-05-15T11:51:55
{ "title": [ -0.03994751, -0.019180298, -0.041900635, -0.027236938, 0.0000042915344, -0.010375977, -0.018356323, -0.011528015, -0.0050468445, 0.026367188, -0.011688232, 0.01424408, -0.009536743, -0.013191223, -0.05493164, 0.0026283264, -0.027420044, ...
http://arxiv.org/abs/2305.08524v2
Measuring Consistency in Text-based Financial Forecasting Models
Linyi Yang, Yingpeng Ma, Yue Zhang
cs.CL, cs.AI, econ.GN, q-fin.EC
Financial forecasting has been an important and active area of machine learning research, as even the most modest advantage in predictive accuracy can be parlayed into significant financial gains. Recent advances in natural language processing (NLP) bring the opportunity to leverage textual data, such as earnings repor...
http://arxiv.org/pdf/2305.08524v2
2023-06-02T05:13:40
2023-05-15T10:32:26
{ "title": [ -0.05831909, 0.01234436, -0.05557251, -0.04296875, -0.0033397675, 0.00969696, -0.05810547, -0.015533447, -0.0059127808, 0.052703857, -0.046722412, -0.0012264252, -0.010925293, 0.008079529, -0.0770874, -0.028930664, -0.011703491, 0.01...
http://arxiv.org/abs/2305.08518v1
Beqi: Revitalize the Senegalese Wolof Language with a Robust Spelling Corrector
Derguene Mbaye, Moussa Diallo
cs.CL, cs.AI
The progress of Natural Language Processing (NLP), although fast in recent years, is not at the same pace for all languages. African languages in particular are still behind and lack automatic processing tools. Some of these tools are very important for the development of these languages but also have an important role...
http://arxiv.org/pdf/2305.08518v1
2023-05-15T10:28:36
2023-05-15T10:28:36
{ "title": [ 0.002450943, -0.021697998, -0.021392822, -0.0053482056, -0.045135498, 0.0013580322, -0.040008545, -0.0063476562, 0.0045814514, 0.049926758, 0.006023407, -0.011650085, 0.0579834, -0.053009033, -0.0181427, 0.013511658, -0.019927979, 0....
http://arxiv.org/abs/2305.08503v5
A Hierarchical Encoding-Decoding Scheme for Abstractive Multi-document Summarization
Chenhui Shen, Liying Cheng, Xuan-Phi Nguyen, Yang You, Lidong Bing
cs.CL
Pre-trained language models (PLMs) have achieved outstanding achievements in abstractive single-document summarization (SDS). However, such benefits may not fully extend to multi-document summarization (MDS), where the handling of cross-document information is more complex. Previous works either design new MDS architec...
http://arxiv.org/pdf/2305.08503v5
2023-11-01T09:18:34
2023-05-15T10:03:31
{ "title": [ -0.014572144, -0.004337311, -0.017456055, -0.0869751, -0.006702423, 0.024627686, -0.069885254, 0.00018334389, 0.01902771, 0.06964111, -0.007583618, 0.021453857, 0.01625061, 0.00093746185, -0.07244873, 0.027160645, -0.0039787292, 0.03...
http://arxiv.org/abs/2305.08502v1
MeeQA: Natural Questions in Meeting Transcripts
Reut Apel, Tom Braude, Amir Kantor, Eyal Kolman
cs.CL, cs.LG
We present MeeQA, a dataset for natural-language question answering over meeting transcripts. It includes real questions asked during meetings by its participants. The dataset contains 48K question-answer pairs, extracted from 422 meeting transcripts, spanning multiple domains. Questions in transcripts pose a special c...
http://arxiv.org/pdf/2305.08502v1
2023-05-15T10:02:47
2023-05-15T10:02:47
{ "title": [ -0.052825928, -0.02319336, -0.06976318, -0.02885437, -0.03112793, -0.010528564, -0.044433594, 0.015090942, 0.015129089, 0.03186035, -0.025817871, 0.0003426075, 0.017059326, -0.025054932, -0.041259766, 0.0054893494, -0.0069236755, 0.0...
http://arxiv.org/abs/2305.08495v1
Similarity-weighted Construction of Contextualized Commonsense Knowledge Graphs for Knowledge-intense Argumentation Tasks
Moritz Plenz, Juri Opitz, Philipp Heinisch, Philipp Cimiano, Anette Frank
cs.CL, cs.DB
Arguments often do not make explicit how a conclusion follows from its premises. To compensate for this lack, we enrich arguments with structured background knowledge to support knowledge-intense argumentation tasks. We present a new unsupervised method for constructing Contextualized Commonsense Knowledge Graphs (CCKG...
http://arxiv.org/pdf/2305.08495v1
2023-05-15T09:52:36
2023-05-15T09:52:36
{ "title": [ 0.028045654, -0.01524353, -0.029327393, -0.08404541, -0.0072250366, -0.002532959, -0.03387451, 0.0024795532, 0.0075035095, 0.021453857, 0.0005888939, 0.02760315, -0.033996582, -0.026733398, -0.032928467, -0.009559631, -0.0073242188, ...
http://arxiv.org/abs/2305.08493v2
Creative Data Generation: A Review Focusing on Text and Poetry
Mohamad Elzohbi, Richard Zhao
cs.CL
The rapid advancement in machine learning has led to a surge in automatic data generation, making it increasingly challenging to differentiate between naturally or human-generated data and machine-generated data. Despite these advancements, the generation of creative data remains a challenge. This paper aims to investi...
http://arxiv.org/pdf/2305.08493v2
2023-06-26T03:50:55
2023-05-15T09:50:15
{ "title": [ -0.026641846, 0.0025691986, -0.039276123, -0.038085938, 0.006954193, -0.0034942627, -0.020904541, -0.008094788, 0.003227234, 0.03387451, -0.0206604, -0.0032157898, -0.035614014, -0.011650085, -0.0736084, -0.016113281, -0.01436615, 0....
http://arxiv.org/abs/2305.08487v1
Taxi1500: A Multilingual Dataset for Text Classification in 1500 Languages
Chunlan Ma, Ayyoob ImaniGooghari, Haotian Ye, Ehsaneddin Asgari, Hinrich Schütze
cs.CL
While natural language processing tools have been developed extensively for some of the world's languages, a significant portion of the world's over 7000 languages are still neglected. One reason for this is that evaluation datasets do not yet cover a wide range of languages, including low-resource and endangered ones....
http://arxiv.org/pdf/2305.08487v1
2023-05-15T09:43:32
2023-05-15T09:43:32
{ "title": [ -0.019180298, -0.008842468, -0.038085938, -0.055419922, 0.012489319, -0.0032100677, -0.048858643, -0.005645752, 0.015838623, 0.030899048, -0.011314392, 0.014801025, -0.010192871, -0.008857727, -0.047943115, 0.0070648193, -0.012390137, ...
http://arxiv.org/abs/2305.08475v2
A Crosslingual Investigation of Conceptualization in 1335 Languages
Yihong Liu, Haotian Ye, Leonie Weissweiler, Philipp Wicke, Renhao Pei, Robert Zangenfeind, Hinrich Schütze
cs.CL
Languages differ in how they divide up the world into concepts and words; e.g., in contrast to English, Swahili has a single concept for `belly' and `womb'. We investigate these differences in conceptualization across 1,335 languages by aligning concepts in a parallel corpus. To this end, we propose Conceptualizer, a m...
http://arxiv.org/pdf/2305.08475v2
2023-05-26T18:29:24
2023-05-15T09:27:34
{ "title": [ -0.004463196, 0.0059013367, -0.021438599, -0.057739258, 0.005001068, -0.007129669, -0.05355835, -0.010292053, -0.0054740906, 0.044311523, 0.00020301342, 0.026031494, -0.009468079, -0.021362305, -0.043395996, 0.019058228, -0.011558533, ...
http://arxiv.org/abs/2305.08473v1
Shared and Private Information Learning in Multimodal Sentiment Analysis with Deep Modal Alignment and Self-supervised Multi-Task Learning
Songning Lai, Xifeng Hu, Yulong Li, Zhaoxia Ren, Zhi Liu, Danmin Miao
cs.CL, cs.CV
Designing an effective representation learning method for multimodal sentiment analysis tasks is a crucial research direction. The challenge lies in learning both shared and private information in a complete modal representation, which is difficult with uniform multimodal labels and a raw feature fusion approach. In th...
http://arxiv.org/pdf/2305.08473v1
2023-05-15T09:24:48
2023-05-15T09:24:48
{ "title": [ -0.014221191, 0.013618469, -0.0496521, -0.048858643, 0.0009880066, -0.020767212, -0.050964355, -0.01524353, -0.0063171387, 0.055999756, -0.010940552, 0.0044670105, 0.009262085, -0.01675415, -0.04373169, 0.0007543564, -0.0044555664, 0...
http://arxiv.org/abs/2305.08455v3
Document Understanding Dataset and Evaluation (DUDE)
Jordy Van Landeghem, Rubén Tito, Łukasz Borchmann, Michał Pietruszka, Paweł Józiak, Rafał Powalski, Dawid Jurkiewicz, Mickaël Coustaty, Bertrand Ackaert, Ernest Valveny, Matthew Blaschko, Sien Moens, Tomasz Stanisławek
cs.CV, cs.CL, cs.LG
We call on the Document AI (DocAI) community to reevaluate current methodologies and embrace the challenge of creating more practically-oriented benchmarks. Document Understanding Dataset and Evaluation (DUDE) seeks to remediate the halted research progress in understanding visually-rich documents (VRDs). We present a ...
http://arxiv.org/pdf/2305.08455v3
2023-09-11T10:36:41
2023-05-15T08:54:32
{ "title": [ -0.0057258606, -0.00605011, -0.06530762, -0.06604004, -0.018585205, -0.02798462, -0.061035156, 0.015655518, -0.0041656494, 0.033416748, -0.04248047, -0.021072388, 0.046783447, -0.035949707, 0.010772705, -0.031280518, 0.011100769, 0.0...
http://arxiv.org/abs/2305.08433v1
EMBRACE: Evaluation and Modifications for Boosting RACE
Mariia Zyrianova, Dmytro Kalpakchi, Johan Boye
cs.CL
When training and evaluating machine reading comprehension models, it is very important to work with high-quality datasets that are also representative of real-world reading comprehension tasks. This requirement includes, for instance, having questions that are based on texts of different genres and require generating ...
http://arxiv.org/pdf/2305.08433v1
2023-05-15T08:21:32
2023-05-15T08:21:32
{ "title": [ -0.0020141602, 0.005027771, -0.040100098, -0.048339844, 0.010353088, -0.0073509216, -0.008773804, -0.01007843, -0.038116455, 0.021820068, -0.029785156, 0.0010356903, -0.015174866, -0.046905518, -0.006767273, -0.017654419, -0.0048675537, ...
http://arxiv.org/abs/2305.08428v1
Legal Extractive Summarization of U.S. Court Opinions
Emmanuel Bauer, Dominik Stammbach, Nianlong Gu, Elliott Ash
cs.CL
This paper tackles the task of legal extractive summarization using a dataset of 430K U.S. court opinions with key passages annotated. According to automated summary quality metrics, the reinforcement-learning-based MemSum model is best and even out-performs transformer-based models. In turn, expert human evaluation sh...
http://arxiv.org/pdf/2305.08428v1
2023-05-15T08:13:00
2023-05-15T08:13:00
{ "title": [ 0.0045280457, -0.020507812, -0.012306213, -0.06378174, -0.04196167, -0.0069084167, -0.034576416, 0.015182495, 0.00086545944, 0.04876709, -0.05718994, 0.029144287, 0.02104187, -0.011917114, -0.014007568, 0.010925293, 0.008857727, 0.02...
http://arxiv.org/abs/2305.08414v1
What's the Meaning of Superhuman Performance in Today's NLU?
Simone Tedeschi, Johan Bos, Thierry Declerck, Jan Hajic, Daniel Hershcovich, Eduard H. Hovy, Alexander Koller, Simon Krek, Steven Schockaert, Rico Sennrich, Ekaterina Shutova, Roberto Navigli
cs.CL, cs.AI
In the last five years, there has been a significant focus in Natural Language Processing (NLP) on developing larger Pretrained Language Models (PLMs) and introducing benchmarks such as SuperGLUE and SQuAD to measure their abilities in language understanding, reasoning, and reading comprehension. These PLMs have achiev...
http://arxiv.org/pdf/2305.08414v1
2023-05-15T07:48:31
2023-05-15T07:48:31
{ "title": [ 0.02909851, 0.0075035095, -0.01689148, -0.019744873, 0.006919861, -0.029556274, 0.046691895, 0.005962372, -0.054718018, 0.04776001, 0.029174805, -0.0051612854, 0.0016431808, -0.035339355, 0.0038070679, -0.0016946793, 0.0054092407, 0....
http://arxiv.org/abs/2305.08391v1
Uncovering the Potential of ChatGPT for Discourse Analysis in Dialogue: An Empirical Study
Yaxin Fan, Feng Jiang
cs.CL
Large Language Models (LLMs) like ChatGPT have proven a great shallow understanding of many traditional NLP tasks, such as translation, summarization, etc. However, its performance on high-level understanding, such as dialogue discourse analysis task that requires a higher level of understanding and reasoning, remains ...
http://arxiv.org/pdf/2305.08391v1
2023-05-15T07:14:41
2023-05-15T07:14:41
{ "title": [ -0.0118637085, -0.01134491, -0.035736084, -0.0413208, 0.006565094, -0.0012016296, -0.029541016, 0.00054883957, -0.018096924, 0.020339966, 0.0131073, 0.019073486, -0.023452759, -0.019927979, -0.047729492, 0.006034851, -0.041046143, 0....
http://arxiv.org/abs/2305.08383v2
The Impact of Incumbent/Opposition Status and Ideological Similitude on Emotions in Political Manifestos
Takumi Nishi
cs.CL
The study involved the analysis of emotion-associated language in the UK Conservative and Labour party general election manifestos between 2000 to 2019. While previous research have shown a general correlation between ideological positioning and overlap of public policies, there are still conflicting results in matters...
http://arxiv.org/pdf/2305.08383v2
2023-09-21T17:18:37
2023-05-15T06:43:44
{ "title": [ 0.04058838, 0.006061554, -0.011108398, -0.030380249, 0.010871887, -0.034851074, 0.0284729, -0.030014038, -0.022003174, 0.03515625, -0.021240234, 0.025482178, -0.0052604675, -0.017929077, -0.029556274, -0.015342712, 0.0026760101, 0.00...
http://arxiv.org/abs/2305.08379v1
TESS: Text-to-Text Self-Conditioned Simplex Diffusion
Rabeeh Karimi Mahabadi, Jaesung Tae, Hamish Ivison, James Henderson, Iz Beltagy, Matthew E. Peters, Arman Cohan
cs.CL, cs.LG
Diffusion models have emerged as a powerful paradigm for generation, obtaining strong performance in various domains with continuous-valued inputs. Despite the promises of fully non-autoregressive text generation, applying diffusion models to natural language remains challenging due to its discrete nature. In this work...
http://arxiv.org/pdf/2305.08379v1
2023-05-15T06:33:45
2023-05-15T06:33:45
{ "title": [ -0.019439697, -0.0034122467, -0.03616333, -0.015853882, -0.01309967, -0.010475159, -0.024871826, -0.0211792, 0.0016555786, 0.039764404, -0.011238098, 0.010246277, 0.020233154, -0.047332764, -0.0072402954, 0.012825012, 0.0059127808, 0...
http://arxiv.org/abs/2305.08377v3
Text Classification via Large Language Models
Xiaofei Sun, Xiaoya Li, Jiwei Li, Fei Wu, Shangwei Guo, Tianwei Zhang, Guoyin Wang
cs.CL
Despite the remarkable success of large-scale Language Models (LLMs) such as GPT-3, their performances still significantly underperform fine-tuned models in the task of text classification. This is due to (1) the lack of reasoning ability in addressing complex linguistic phenomena (e.g., intensification, contrast, iron...
http://arxiv.org/pdf/2305.08377v3
2023-10-09T15:52:30
2023-05-15T06:24:45
{ "title": [ -0.025344849, 0.00075531006, -0.032196045, -0.04699707, 0.005302429, 0.0012674332, -0.041625977, -0.012580872, 0.013824463, 0.021408081, -0.024612427, 0.0023422241, -0.014251709, -0.0070495605, -0.046813965, 0.00015234947, -0.018447876, ...
http://arxiv.org/abs/2305.08372v1
A Novel Framework for Multimodal Named Entity Recognition with Multi-level Alignments
Peipei Liu, Hong Li, Yimo Ren, Jie Liu, Shuaizong Si, Hongsong Zhu, Limin Sun
cs.CL, cs.MM
Mining structured knowledge from tweets using named entity recognition (NER) can be beneficial for many downstream applications such as recommendation and intention under standing. With tweet posts tending to be multimodal, multimodal named entity recognition (MNER) has attracted more attention. In this paper, we propo...
http://arxiv.org/pdf/2305.08372v1
2023-05-15T06:14:36
2023-05-15T06:14:36
{ "title": [ -0.01802063, 0.0073623657, -0.06536865, -0.080444336, 0.0287323, -0.0013589859, -0.05291748, 0.017166138, 0.005580902, 0.028747559, -0.005569458, 0.01259613, 0.007583618, -0.013916016, -0.035247803, 0.012512207, -0.010795593, 0.03802...
http://arxiv.org/abs/2305.08371v2
SuperDialseg: A Large-scale Dataset for Supervised Dialogue Segmentation
Junfeng Jiang, Chengzhang Dong, Sadao Kurohashi, Akiko Aizawa
cs.CL
Dialogue segmentation is a crucial task for dialogue systems allowing a better understanding of conversational texts. Despite recent progress in unsupervised dialogue segmentation methods, their performances are limited by the lack of explicit supervised signals for training. Furthermore, the precise definition of segm...
http://arxiv.org/pdf/2305.08371v2
2023-10-15T15:03:17
2023-05-15T06:08:01
{ "title": [ -0.0065193176, -0.0021896362, -0.05142212, -0.068115234, 0.010597229, -0.005256653, -0.04788208, 0.0096588135, -0.004486084, 0.010055542, 0.002319336, 0.007080078, -0.023895264, -0.021697998, -0.023376465, -0.0049934387, -0.0077438354, ...
http://arxiv.org/abs/2305.10445v2
Memorization for Good: Encryption with Autoregressive Language Models
Samuel Stevens, Yu Su
cs.CL, cs.AI
Over-parameterized neural language models (LMs) can memorize and recite long sequences of training data. While such memorization is normally associated with undesired properties such as overfitting and information leaking, our work casts memorization as an unexplored capability of LMs. We propose the first symmetric en...
http://arxiv.org/pdf/2305.10445v2
2023-10-13T18:25:11
2023-05-15T05:42:34
{ "title": [ -0.045928955, -0.000007390976, -0.047729492, -0.031829834, -0.0018939972, -0.005695343, -0.039093018, 0.0012311935, -0.013214111, 0.04949951, 0.007167816, -0.0009741783, 0.013145447, -0.02218628, -0.05029297, 0.003730774, -0.002319336, ...
http://arxiv.org/abs/2305.08348v2
Coreference-aware Double-channel Attention Network for Multi-party Dialogue Reading Comprehension
Yanling Li, Bowei Zou, Yifan Fan, Mengxing Dong, Yu Hong
cs.CL
We tackle Multi-party Dialogue Reading Comprehension (abbr., MDRC). MDRC stands for an extractive reading comprehension task grounded on a batch of dialogues among multiple interlocutors. It is challenging due to the requirement of understanding cross-utterance contexts and relationships in a multi-turn multi-party con...
http://arxiv.org/pdf/2305.08348v2
2023-05-22T04:42:40
2023-05-15T05:01:29
{ "title": [ -0.016799927, -0.019378662, -0.0184021, -0.049987793, -0.033111572, 0.00072336197, -0.05316162, 0.016921997, 0.017410278, 0.047210693, 0.0069503784, 0.008842468, 0.026351929, -0.015991211, -0.05166626, 0.029327393, -0.016281128, 0.02...
http://arxiv.org/abs/2305.08347v1
KEPR: Knowledge Enhancement and Plausibility Ranking for Generative Commonsense Question Answering
Zhifeng Li, Bowei Zou, Yifan Fan, Yu Hong
cs.CL, cs.AI
Generative commonsense question answering (GenCQA) is a task of automatically generating a list of answers given a question. The answer list is required to cover all reasonable answers. This presents the considerable challenges of producing diverse answers and ranking them properly. Incorporating a variety of closely-r...
http://arxiv.org/pdf/2305.08347v1
2023-05-15T04:58:37
2023-05-15T04:58:37
{ "title": [ 0.00667572, -0.01991272, -0.021957397, -0.053710938, -0.010978699, -0.0023841858, -0.030883789, -0.013687134, 0.04244995, 0.00484848, -0.024108887, 0.0082473755, -0.01676941, -0.020050049, -0.042907715, 0.0048103333, -0.016677856, 0....
http://arxiv.org/abs/2305.08339v3
Assessing the potential of AI-assisted pragmatic annotation: The case of apologies
Danni Yu, Luyang Li, Hang Su, Matteo Fuoli
cs.CL, cs.AI
Certain forms of linguistic annotation, like part of speech and semantic tagging, can be automated with high accuracy. However, manual annotation is still necessary for complex pragmatic and discursive features that lack a direct mapping to lexical forms. This manual process is time-consuming and error-prone, limiting ...
http://arxiv.org/pdf/2305.08339v3
2023-09-15T09:55:47
2023-05-15T04:10:13
{ "title": [ -0.0063171387, -0.010307312, -0.0054893494, -0.031280518, -0.034362793, -0.0104599, -0.0043563843, -0.002866745, 0.009750366, 0.033050537, -0.0340271, -0.021102905, 0.015838623, -0.015014648, -0.07507324, -0.013748169, -0.025680542, ...
http://arxiv.org/abs/2305.08322v3
C-Eval: A Multi-Level Multi-Discipline Chinese Evaluation Suite for Foundation Models
Yuzhen Huang, Yuzhuo Bai, Zhihao Zhu, Junlei Zhang, Jinghan Zhang, Tangjun Su, Junteng Liu, Chuancheng Lv, Yikai Zhang, Jiayi Lei, Yao Fu, Maosong Sun, Junxian He
cs.CL
New NLP benchmarks are urgently needed to align with the rapid development of large language models (LLMs). We present C-Eval, the first comprehensive Chinese evaluation suite designed to assess advanced knowledge and reasoning abilities of foundation models in a Chinese context. C-Eval comprises multiple-choice questi...
http://arxiv.org/pdf/2305.08322v3
2023-11-06T13:24:16
2023-05-15T03:20:19
{ "title": [ -0.053894043, 0.0058403015, -0.03955078, -0.04537964, 0.0026378632, 0.0085372925, -0.04034424, -0.017425537, 0.0017061234, 0.010070801, 0.009513855, -0.01374054, -0.027023315, -0.05230713, -0.029785156, 0.004840851, -0.037841797, 0.0...
http://arxiv.org/abs/2305.16328v1
Semantic Composition in Visually Grounded Language Models
Rohan Pandey
cs.CL, cs.LG
What is sentence meaning and its ideal representation? Much of the expressive power of human language derives from semantic composition, the mind's ability to represent meaning hierarchically & relationally over constituents. At the same time, much sentential meaning is outside the text and requires grounding in sensor...
http://arxiv.org/pdf/2305.16328v1
2023-05-15T03:19:42
2023-05-15T03:19:42
{ "title": [ -0.019332886, 0.0036392212, -0.024856567, -0.08337402, -0.0061683655, 0.014755249, -0.08508301, -0.0096206665, 0.024032593, 0.045684814, -0.012580872, -0.0035762787, 0.007133484, -0.009689331, -0.07043457, 0.0014953613, -0.014518738, ...
http://arxiv.org/abs/2305.08300v1
"Nothing Abnormal": Disambiguating Medical Reports via Contrastive Knowledge Infusion
Zexue He, An Yan, Amilcare Gentili, Julian McAuley, Chun-Nan Hsu
cs.CL
Sharing medical reports is essential for patient-centered care. A recent line of work has focused on automatically generating reports with NLP methods. However, different audiences have different purposes when writing/reading medical reports -- for example, healthcare professionals care more about pathology, whereas pa...
http://arxiv.org/pdf/2305.08300v1
2023-05-15T02:01:20
2023-05-15T02:01:20
{ "title": [ 0.0032310486, -0.006111145, -0.034423828, -0.07965088, 0.004085541, -0.0066375732, -0.028305054, -0.0062561035, 0.0087890625, 0.040374756, -0.02267456, 0.03564453, 0.00032901764, -0.007255554, -0.048828125, 0.005466461, -0.0032844543, ...
http://arxiv.org/abs/2305.08298v1
Symbol tuning improves in-context learning in language models
Jerry Wei, Le Hou, Andrew Lampinen, Xiangning Chen, Da Huang, Yi Tay, Xinyun Chen, Yifeng Lu, Denny Zhou, Tengyu Ma, Quoc V. Le
cs.CL
We present symbol tuning - finetuning language models on in-context input-label pairs where natural language labels (e.g., "positive/negative sentiment") are replaced with arbitrary symbols (e.g., "foo/bar"). Symbol tuning leverages the intuition that when a model cannot use instructions or natural language labels to f...
http://arxiv.org/pdf/2305.08298v1
2023-05-15T01:59:58
2023-05-15T01:59:58
{ "title": [ -0.0066719055, -0.0053138733, -0.008483887, -0.030014038, -0.031982422, -0.011184692, -0.029129028, -0.002216339, -0.02178955, 0.038513184, -0.016738892, 0.0038776398, 0.0060310364, -0.02268982, -0.031219482, 0.0005493164, 0.0011482239, ...
http://arxiv.org/abs/2305.08291v1
Large Language Model Guided Tree-of-Thought
Jieyi Long
cs.AI, cs.CL, cs.CV, cs.LG, cs.NE
In this paper, we introduce the Tree-of-Thought (ToT) framework, a novel approach aimed at improving the problem-solving capabilities of auto-regressive large language models (LLMs). The ToT technique is inspired by the human mind's approach for solving complex reasoning tasks through trial and error. In this process, ...
http://arxiv.org/pdf/2305.08291v1
2023-05-15T01:18:23
2023-05-15T01:18:23
{ "title": [ 0.0030441284, -0.029464722, -0.010528564, -0.0446167, -0.07775879, 0.0033073425, -0.058166504, 0.020767212, 0.012626648, 0.029891968, -0.04046631, -0.023254395, 0.053497314, -0.049438477, 0.00422287, 0.0022144318, 0.008918762, 0.0295...
http://arxiv.org/abs/2305.08285v3
Parameter-Efficient Fine-Tuning with Layer Pruning on Free-Text Sequence-to-Sequence Modeling
Yunqi Zhu, Xuebing Yang, Yuanyuan Wu, Wensheng Zhang
cs.CL, cs.AI
The increasing size of language models raises great research interests in parameter-efficient fine-tuning such as LoRA that freezes the pre-trained model, and injects small-scale trainable parameters for multiple downstream tasks (e.g., summarization, question answering and translation). To further enhance the efficien...
http://arxiv.org/pdf/2305.08285v3
2023-05-19T01:29:08
2023-05-15T00:21:08
{ "title": [ -0.0115356445, -0.010017395, 0.00617218, -0.096069336, -0.0096588135, 0.027954102, -0.059387207, 0.018829346, 0.0055656433, 0.04284668, 0.0025863647, 0.011978149, 0.01889038, -0.01576233, -0.049713135, 0.012542725, -0.0077667236, 0.0...
http://arxiv.org/abs/2305.08283v3
From Pretraining Data to Language Models to Downstream Tasks: Tracking the Trails of Political Biases Leading to Unfair NLP Models
Shangbin Feng, Chan Young Park, Yuhan Liu, Yulia Tsvetkov
cs.CL
Language models (LMs) are pretrained on diverse data sources, including news, discussion forums, books, and online encyclopedias. A significant portion of this data includes opinions and perspectives which, on one hand, celebrate democracy and diversity of ideas, and on the other hand are inherently socially biased. Ou...
http://arxiv.org/pdf/2305.08283v3
2023-07-06T00:40:53
2023-05-15T00:06:30
{ "title": [ 0.010559082, -0.012329102, -0.016159058, -0.040649414, -0.02268982, -0.022903442, -0.017623901, -0.020812988, -0.029418945, 0.012229919, -0.035095215, 0.02458191, -0.017166138, -0.018508911, -0.010246277, -0.03010559, -0.02960205, 0....
http://arxiv.org/abs/2305.08281v2
FactKB: Generalizable Factuality Evaluation using Language Models Enhanced with Factual Knowledge
Shangbin Feng, Vidhisha Balachandran, Yuyang Bai, Yulia Tsvetkov
cs.CL
Evaluating the factual consistency of automatically generated summaries is essential for the progress and adoption of reliable summarization systems. Despite recent advances, existing factuality evaluation models are not robust, being especially prone to entity and relation errors in new domains. We propose FactKB, a s...
http://arxiv.org/pdf/2305.08281v2
2023-10-18T23:36:43
2023-05-14T23:58:05
{ "title": [ -0.02468872, -0.0033416748, -0.054473877, -0.07128906, -0.0032138824, -0.002937317, -0.06427002, -0.015625, 0.020523071, 0.037261963, -0.00548172, 0.000053048134, 0.021652222, -0.018218994, -0.053222656, 0.022506714, -0.024993896, 0....
http://arxiv.org/abs/2305.08271v1
$SmartProbe$: A Virtual Moderator for Market Research Surveys
Josh Seltzer, Jiahua Pan, Kathy Cheng, Yuxiao Sun, Santosh Kolagati, Jimmy Lin, Shi Zong
cs.CL
Market research surveys are a powerful methodology for understanding consumer perspectives at scale, but are limited by depth of understanding and insights. A virtual moderator can introduce elements of qualitative research into surveys, developing a rapport with survey participants and dynamically asking probing quest...
http://arxiv.org/pdf/2305.08271v1
2023-05-14T22:36:08
2023-05-14T22:36:08
{ "title": [ -0.01927185, -0.007221222, -0.036254883, -0.010871887, 0.011070251, -0.019165039, 0.0074806213, -0.034942627, -0.004753113, 0.032836914, -0.008811951, -0.0020065308, -0.010147095, -0.03378296, -0.059814453, -0.0058250427, -0.0036468506, ...
http://arxiv.org/abs/2305.08264v1
MatSci-NLP: Evaluating Scientific Language Models on Materials Science Language Tasks Using Text-to-Schema Modeling
Yu Song, Santiago Miret, Bang Liu
cs.CL, cond-mat.mtrl-sci, cs.AI
We present MatSci-NLP, a natural language benchmark for evaluating the performance of natural language processing (NLP) models on materials science text. We construct the benchmark from publicly available materials science text data to encompass seven different NLP tasks, including conventional NLP tasks like named ent...
http://arxiv.org/pdf/2305.08264v1
2023-05-14T22:01:24
2023-05-14T22:01:24
{ "title": [ -0.06829834, -0.021636963, -0.067871094, -0.022338867, -0.024261475, -0.005393982, -0.05996704, 0.01146698, 0.01651001, 0.011550903, 0.0075645447, -0.019836426, 0.010375977, -0.052001953, -0.019882202, 0.017868042, -0.035064697, 0.04...
http://arxiv.org/abs/2305.08246v1
Learning Non-linguistic Skills without Sacrificing Linguistic Proficiency
Mandar Sharma, Nikhil Muralidhar, Naren Ramakrishnan
cs.CL, cs.AI, cs.LG
The field of Math-NLP has witnessed significant growth in recent years, motivated by the desire to expand LLM performance to the learning of non-linguistic notions (numerals, and subsequently, arithmetic reasoning). However, non-linguistic skill injection typically comes at a cost for LLMs: it leads to catastrophic for...
http://arxiv.org/pdf/2305.08246v1
2023-05-14T20:57:11
2023-05-14T20:57:11
{ "title": [ 0.027359009, -0.019577026, -0.03479004, -0.016937256, -0.041381836, -0.034973145, -0.011428833, 0.0004453659, -0.025741577, 0.019882202, -0.00084352493, -0.019500732, -0.014762878, -0.04937744, 0.030929565, -0.03213501, 0.004966736, ...
http://arxiv.org/abs/2305.08227v1
DeepFilterNet: Perceptually Motivated Real-Time Speech Enhancement
Hendrik Schröter, Tobias Rosenkranz, Alberto N. Escalante-B., Andreas Maier
eess.AS, cs.CL, cs.SD
Multi-frame algorithms for single-channel speech enhancement are able to take advantage from short-time correlations within the speech signal. Deep Filtering (DF) was proposed to directly estimate a complex filter in frequency domain to take advantage of these correlations. In this work, we present a real-time speech e...
http://arxiv.org/pdf/2305.08227v1
2023-05-14T19:09:35
2023-05-14T19:09:35
{ "title": [ 0.01134491, 0.003370285, -0.051879883, -0.079711914, 0.027420044, -0.012832642, -0.024841309, -0.0014448166, 0.004005432, 0.04510498, -0.0032806396, -0.0015907288, 0.032409668, -0.025817871, -0.040618896, 0.016433716, -0.0143585205, ...
http://arxiv.org/abs/2305.08208v2
Learning to Generalize for Cross-domain QA
Yingjie Niu, Linyi Yang, Ruihai Dong, Yue Zhang
cs.CL, cs.AI
There have been growing concerns regarding the out-of-domain generalization ability of natural language processing (NLP) models, particularly in question-answering (QA) tasks. Current synthesized data augmentation methods for QA are hampered by increased training costs. To address this issue, we propose a novel approac...
http://arxiv.org/pdf/2305.08208v2
2023-05-18T10:46:20
2023-05-14T17:53:54
{ "title": [ -0.008628845, 0.019851685, -0.044403076, -0.05142212, 0.025878906, -0.019241333, -0.020614624, -0.010612488, -0.011650085, 0.036315918, -0.023071289, 0.012908936, -0.000021398067, -0.0056915283, -0.031555176, -0.0022602081, -0.02368164, ...
http://arxiv.org/abs/2305.08200v1
A Cognitive Stimulation Dialogue System with Multi-source Knowledge Fusion for Elders with Cognitive Impairment
Jiyue Jiang, Sheng Wang, Qintong Li, Lingpeng Kong, Chuan Wu
cs.CL
When communicating with elders with cognitive impairment, cognitive stimulation (CS) help to maintain the cognitive health of elders. Data sparsity is the main challenge in building CS-based dialogue systems, particularly in the Chinese language. To fill this gap, we construct a Chinese CS conversation (CSConv) dataset...
http://arxiv.org/pdf/2305.08200v1
2023-05-14T16:52:20
2023-05-14T16:52:20
{ "title": [ 0.013259888, 0.0027370453, -0.044952393, -0.09118652, 0.028457642, -0.0057029724, -0.038330078, 0.012969971, 0.014923096, 0.03277588, -0.044891357, -0.017700195, 0.007572174, -0.017318726, -0.06262207, -0.028320312, -0.0055007935, 0....
http://arxiv.org/abs/2305.08195v2
Learning to Simulate Natural Language Feedback for Interactive Semantic Parsing
Hao Yan, Saurabh Srivastava, Yintao Tai, Sida I. Wang, Wen-tau Yih, Ziyu Yao
cs.CL
Interactive semantic parsing based on natural language (NL) feedback, where users provide feedback to correct the parser mistakes, has emerged as a more practical scenario than the traditional one-shot semantic parsing. However, prior work has heavily relied on human-annotated feedback data to train the interactive sem...
http://arxiv.org/pdf/2305.08195v2
2023-06-04T21:05:26
2023-05-14T16:20:09
{ "title": [ -0.033111572, -0.0003271103, -0.015617371, -0.0619812, -0.011833191, 0.017196655, -0.0725708, -0.019317627, -0.019546509, 0.017288208, -0.04449463, 0.022155762, -0.028121948, 0.00077724457, -0.033569336, -0.026489258, -0.009880066, 0...
http://arxiv.org/abs/2305.08187v1
CroSentiNews 2.0: A Sentence-Level News Sentiment Corpus
Gaurish Thakkar, Nives Mikelic Preradović, Marko Tadić
cs.CL
This article presents a sentence-level sentiment dataset for the Croatian news domain. In addition to the 3K annotated texts already present, our dataset contains 14.5K annotated sentence occurrences that have been tagged with 5 classes. We provide baseline scores in addition to the annotation process and inter-annotat...
http://arxiv.org/pdf/2305.08187v1
2023-05-14T15:53:54
2023-05-14T15:53:54
{ "title": [ -0.01448822, -0.011878967, -0.035125732, -0.053985596, 0.015350342, -0.0013723373, -0.014266968, -0.005874634, 0.025939941, 0.03665161, -0.03717041, 0.008705139, 0.0026245117, 0.012962341, -0.061767578, 0.0074501038, -0.012573242, 0....
http://arxiv.org/abs/2305.08173v1
Croatian Film Review Dataset (Cro-FiReDa): A Sentiment Annotated Dataset of Film Reviews
Gaurish Thakkar, Nives Mikelic Preradovic, Marko Tadić
cs.CL
This paper introduces Cro-FiReDa, a sentiment-annotated dataset for Croatian in the domain of movie reviews. The dataset, which contains over 10,000 sentences, has been annotated at the sentence level. In addition to presenting the overall annotation process, we also present benchmark results based on the transformer-b...
http://arxiv.org/pdf/2305.08173v1
2023-05-14T14:46:12
2023-05-14T14:46:12
{ "title": [ -0.009735107, -0.008354187, -0.04434204, 0.009857178, -0.013572693, -0.024414062, 0.004299164, -0.012840271, 0.00070142746, 0.050689697, -0.07507324, 0.016143799, 0.052093506, 0.0038986206, -0.010688782, -0.0073242188, -0.001159668, ...
http://arxiv.org/abs/2305.08152v1
STORYWARS: A Dataset and Instruction Tuning Baselines for Collaborative Story Understanding and Generation
Yulun Du, Lydia Chilton
cs.CL
Collaborative stories, which are texts created through the collaborative efforts of multiple authors with different writing styles and intentions, pose unique challenges for NLP models. Understanding and generating such stories remains an underexplored area due to the lack of open-domain corpora. To address this, we in...
http://arxiv.org/pdf/2305.08152v1
2023-05-14T13:09:27
2023-05-14T13:09:27
{ "title": [ -0.053985596, -0.023529053, -0.04486084, -0.049865723, -0.0211792, 0.0029659271, -0.049102783, 0.01285553, -0.0072631836, 0.010360718, -0.027908325, 0.011680603, -0.032562256, -0.0067214966, -0.0335083, -0.030303955, -0.024429321, 0....
http://arxiv.org/abs/2305.08146v1
ParaLS: Lexical Substitution via Pretrained Paraphraser
Jipeng Qiang, Kang Liu, Yun Li, Yunhao Yuan, Yi Zhu
cs.CL, cs.AI
Lexical substitution (LS) aims at finding appropriate substitutes for a target word in a sentence. Recently, LS methods based on pretrained language models have made remarkable progress, generating potential substitutes for a target word through analysis of its contextual surroundings. However, these methods tend to ov...
http://arxiv.org/pdf/2305.08146v1
2023-05-14T12:49:16
2023-05-14T12:49:16
{ "title": [ -0.013305664, -0.012878418, -0.019927979, -0.085510254, -0.007270813, 0.021011353, -0.06463623, -0.009674072, -0.02835083, 0.023910522, -0.0076789856, 0.0037460327, 0.036621094, -0.029296875, 0.0014667511, 0.0016756058, -0.020843506, ...
http://arxiv.org/abs/2305.08135v2
Distinguish Before Answer: Generating Contrastive Explanation as Knowledge for Commonsense Question Answering
Qianglong Chen, Guohai Xu, Ming Yan, Ji Zhang, Fei Huang, Luo Si, Yin Zhang
cs.CL, cs.AI
Existing knowledge-enhanced methods have achieved remarkable results in certain QA tasks via obtaining diverse knowledge from different knowledge bases. However, limited by the properties of retrieved knowledge, they still have trouble benefiting from both the knowledge relevance and distinguishment simultaneously. To ...
http://arxiv.org/pdf/2305.08135v2
2023-05-21T15:07:23
2023-05-14T12:12:24
{ "title": [ 0.01550293, -0.015731812, -0.021194458, -0.069885254, -0.0012626648, -0.013053894, -0.001616478, -0.0050621033, 0.015945435, -0.00022435188, -0.024490356, 0.023712158, -0.031066895, -0.015357971, -0.03765869, -0.0072898865, -0.007881165,...
http://arxiv.org/abs/2305.08099v3
Self-supervised Neural Factor Analysis for Disentangling Utterance-level Speech Representations
Weiwei Lin, Chenhang He, Man-Wai Mak, Youzhi Tu
cs.SD, cs.CL, cs.LG, eess.AS
Self-supervised learning (SSL) speech models such as wav2vec and HuBERT have demonstrated state-of-the-art performance on automatic speech recognition (ASR) and proved to be extremely useful in low label-resource settings. However, the success of SSL models has yet to transfer to utterance-level tasks such as speaker, ...
http://arxiv.org/pdf/2305.08099v3
2023-10-04T12:15:56
2023-05-14T08:26:24
{ "title": [ -0.001748085, 0.016357422, -0.049926758, -0.06945801, 0.003850937, -0.0065345764, -0.055023193, 0.014076233, 0.0025100708, 0.037200928, 0.014846802, -0.009994507, 0.0070648193, -0.040740967, -0.046142578, 0.010032654, -0.01966858, 0....
http://arxiv.org/abs/2305.08096v1
Towards Understanding and Improving Knowledge Distillation for Neural Machine Translation
Songming Zhang, Yunlong Liang, Shuaibo Wang, Wenjuan Han, Jian Liu, Jinan Xu, Yufeng Chen
cs.CL, cs.AI
Knowledge distillation (KD) is a promising technique for model compression in neural machine translation. However, where the knowledge hides in KD is still not clear, which may hinder the development of KD. In this work, we first unravel this mystery from an empirical perspective and show that the knowledge comes from ...
http://arxiv.org/pdf/2305.08096v1
2023-05-14T08:23:03
2023-05-14T08:23:03
{ "title": [ -0.024780273, -0.0038814545, -0.0362854, -0.06707764, -0.0020122528, 0.005672455, -0.06317139, 0.0104599, 0.013328552, 0.04043579, -0.02973938, -0.01499176, 0.015274048, -0.0039367676, -0.053833008, 0.00093078613, 0.00077724457, 0.00...
http://arxiv.org/abs/2305.08883v1
Watermarking Text Generated by Black-Box Language Models
Xi Yang, Kejiang Chen, Weiming Zhang, Chang Liu, Yuang Qi, Jie Zhang, Han Fang, Nenghai Yu
cs.CL, cs.AI
LLMs now exhibit human-like skills in various fields, leading to worries about misuse. Thus, detecting generated text is crucial. However, passive detection methods are stuck in domain specificity and limited adversarial robustness. To achieve reliable detection, a watermark-based method was proposed for white-box LLMs...
http://arxiv.org/pdf/2305.08883v1
2023-05-14T07:37:33
2023-05-14T07:37:33
{ "title": [ -0.02305603, 0.011436462, -0.030761719, -0.043060303, 0.0056762695, -0.0023975372, -0.03604126, -0.026977539, -0.017990112, 0.04638672, -0.0029716492, 0.015426636, 0.013671875, -0.0138168335, -0.03930664, 0.00667572, -0.019088745, 0....
http://arxiv.org/abs/2305.08088v1
Make Prompt-based Black-Box Tuning Colorful: Boosting Model Generalization from Three Orthogonal Perspectives
Qiushi Sun, Chengcheng Han, Nuo Chen, Renyu Zhu, Jingyang Gong, Xiang Li, Ming Gao
cs.CL, cs.AI
Large language models (LLMs) have shown increasing power on various natural language processing (NLP) tasks. However, tuning these models for downstream tasks usually needs exorbitant costs or is unavailable due to commercial considerations. Recently, black-box tuning has been proposed to address this problem by optimi...
http://arxiv.org/pdf/2305.08088v1
2023-05-14T07:33:59
2023-05-14T07:33:59
{ "title": [ 0.015602112, 0.013427734, -0.023406982, -0.053344727, -0.019439697, -0.0072250366, -0.07470703, -0.022323608, -0.049346924, 0.026565552, -0.03387451, 0.029724121, -0.018325806, -0.006576538, -0.016036987, -0.04159546, 0.010482788, 0....
http://arxiv.org/abs/2305.08067v1
Improving End-to-End SLU performance with Prosodic Attention and Distillation
Shangeth Rajaa
cs.CL, cs.LG, cs.SD, eess.AS
Most End-to-End SLU methods depend on the pretrained ASR or language model features for intent prediction. However, other essential information in speech, such as prosody, is often ignored. Recent research has shown improved results in classifying dialogue acts by incorporating prosodic information. The margins of impr...
http://arxiv.org/pdf/2305.08067v1
2023-05-14T04:38:20
2023-05-14T04:38:20
{ "title": [ 0.021621704, 0.028259277, -0.045440674, -0.08648682, 0.004272461, -0.011749268, -0.057769775, 0.012687683, -0.003774643, 0.049346924, 0.0038814545, -0.029220581, 0.048736572, -0.046569824, -0.02319336, 0.006603241, -0.005596161, -0.0...
http://arxiv.org/abs/2305.08059v1
Semantic-aware Dynamic Retrospective-Prospective Reasoning for Event-level Video Question Answering
Chenyang Lyu, Tianbo Ji, Yvette Graham, Jennifer Foster
cs.CV, cs.AI, cs.CL
Event-Level Video Question Answering (EVQA) requires complex reasoning across video events to obtain the visual information needed to provide optimal answers. However, despite significant progress in model performance, few studies have focused on using the explicit semantic connections between the question and visual i...
http://arxiv.org/pdf/2305.08059v1
2023-05-14T03:57:11
2023-05-14T03:57:11
{ "title": [ 0.027816772, 0.0055732727, -0.041229248, -0.13427734, 0.016342163, 0.0036811829, -0.045715332, 0.034118652, 0.019638062, 0.021469116, -0.012504578, -0.011047363, 0.0004043579, -0.022018433, -0.04421997, -0.013641357, 0.0178833, 0.008...
http://arxiv.org/abs/2305.08010v2
ProKnow: Process Knowledge for Safety Constrained and Explainable Question Generation for Mental Health Diagnostic Assistance
Kaushik Roy, Manas Gaur, Misagh Soltani, Vipula Rawte, Ashwin Kalyan, Amit Sheth
cs.CL
Current Virtual Mental Health Assistants (VMHAs) provide counseling and suggestive care. They refrain from patient diagnostic assistance because they lack training in safety-constrained and specialized clinical process knowledge. In this work, we define Proknow as an ordered set of information that maps to evidence-bas...
http://arxiv.org/pdf/2305.08010v2
2023-06-01T18:33:33
2023-05-13T21:31:02
{ "title": [ -0.0026435852, -0.015960693, -0.038482666, -0.086120605, -0.0014448166, 0.010574341, -0.05545044, 0.014663696, 0.0072288513, -0.0022830963, 0.0043411255, -0.0075149536, 0.023712158, -0.06555176, 0.009223938, 0.006839752, 0.017654419, ...
http://arxiv.org/abs/2305.08005v1
Beyond the Safeguards: Exploring the Security Risks of ChatGPT
Erik Derner, Kristina Batistič
cs.CR, cs.AI, cs.CL, cs.CY, cs.HC
The increasing popularity of large language models (LLMs) such as ChatGPT has led to growing concerns about their safety, security risks, and ethical implications. This paper aims to provide an overview of the different types of security risks associated with ChatGPT, including malicious text and code generation, priva...
http://arxiv.org/pdf/2305.08005v1
2023-05-13T21:01:14
2023-05-13T21:01:14
{ "title": [ -0.0076789856, -0.0070381165, -0.021209717, -0.0340271, 0.014419556, -0.000018656254, -0.008781433, -0.012466431, -0.043426514, 0.025131226, 0.0017604828, 0.03262329, -0.019180298, -0.022827148, -0.021636963, 0.0022773743, -0.018722534, ...
http://arxiv.org/abs/2305.07991v2
Multilingual Previously Fact-Checked Claim Retrieval
Matúš Pikuliak, Ivan Srba, Robert Moro, Timo Hromadka, Timotej Smolen, Martin Melisek, Ivan Vykopal, Jakub Simko, Juraj Podrouzek, Maria Bielikova
cs.CL
Fact-checkers are often hampered by the sheer amount of online content that needs to be fact-checked. NLP can help them by retrieving already existing fact-checks relevant to the content being investigated. This paper introduces a new multilingual dataset -- MultiClaim -- for previously fact-checked claim retrieval. We...
http://arxiv.org/pdf/2305.07991v2
2023-10-13T20:47:57
2023-05-13T20:00:18
{ "title": [ 0.024124146, -0.0018920898, -0.022827148, -0.07128906, -0.01133728, -0.015594482, -0.04034424, -0.009483337, -0.022567749, 0.060333252, 0.0031833649, 0.005428314, 0.031280518, -0.024230957, -0.010818481, 0.009346008, -0.008110046, 0....
http://arxiv.org/abs/2305.07988v2
Reconstruct Before Summarize: An Efficient Two-Step Framework for Condensing and Summarizing Meeting Transcripts
Haochen Tan, Han Wu, Wei Shao, Xinyun Zhang, Mingjie Zhan, Zhaohui Hou, Ding Liang, Linqi Song
cs.CL
Meetings typically involve multiple participants and lengthy conversations, resulting in redundant and trivial content. To overcome these challenges, we propose a two-step framework, Reconstruct before Summarize (RbS), for effective and efficient meeting summarization. RbS first leverages a self-supervised paradigm to ...
http://arxiv.org/pdf/2305.07988v2
2023-10-22T17:42:44
2023-05-13T19:54:46
{ "title": [ -0.026519775, -0.0107421875, -0.066467285, -0.10571289, -0.012771606, 0.008232117, -0.1104126, 0.0076065063, -0.028686523, 0.05053711, -0.021209717, 0.040283203, 0.021194458, 0.010696411, -0.024597168, -0.006969452, -0.01272583, 0.02...
http://arxiv.org/abs/2305.07984v2
SCENE: Self-Labeled Counterfactuals for Extrapolating to Negative Examples
Deqing Fu, Ameya Godbole, Robin Jia
cs.CL
Detecting negatives (such as non-entailment relationships, unanswerable questions, and false claims) is an important and challenging aspect of many natural language understanding tasks. Though manually collecting challenging negative examples can help models detect them, it is both costly and domain-specific. In this w...
http://arxiv.org/pdf/2305.07984v2
2023-11-01T03:02:03
2023-05-13T19:30:58
{ "title": [ 0.08959961, -0.0005412102, 0.024520874, -0.012931824, -0.02330017, -0.024261475, 0.015686035, -0.06304932, -0.009750366, -0.0052604675, -0.06689453, 0.0068855286, -0.0317688, -0.032592773, 0.009010315, -0.055358887, 0.0017318726, 0.0...
http://arxiv.org/abs/2305.07982v2
Zero-shot Faithful Factual Error Correction
Kung-Hsiang Huang, Hou Pong Chan, Heng Ji
cs.CL
Faithfully correcting factual errors is critical for maintaining the integrity of textual knowledge bases and preventing hallucinations in sequence-to-sequence models. Drawing on humans' ability to identify and correct factual errors, we present a zero-shot framework that formulates questions about input claims, looks ...
http://arxiv.org/pdf/2305.07982v2
2023-05-27T15:38:29
2023-05-13T18:55:20
{ "title": [ 0.048950195, 0.001786232, -0.03555298, -0.038513184, -0.032104492, -0.039642334, -0.01676941, -0.002878189, -0.00182724, 0.070495605, -0.007835388, -0.014755249, 0.055786133, -0.030136108, -0.02027893, 0.023132324, -0.0005226135, 0.0...
http://arxiv.org/abs/2305.07972v1
Trillion Dollar Words: A New Financial Dataset, Task & Market Analysis
Agam Shah, Suvan Paturi, Sudheer Chava
cs.CL, q-fin.CP
Monetary policy pronouncements by Federal Open Market Committee (FOMC) are a major driver of financial market returns. We construct the largest tokenized and annotated dataset of FOMC speeches, meeting minutes, and press conference transcripts in order to understand how monetary policy influences financial markets. In ...
http://arxiv.org/pdf/2305.07972v1
2023-05-13T17:32:39
2023-05-13T17:32:39
{ "title": [ -0.015159607, -0.013671875, -0.07550049, -0.04598999, 0.008232117, -0.043182373, -0.021194458, 0.009590149, -0.040649414, 0.028305054, -0.051452637, 0.024642944, 0.010757446, -0.019180298, 0.00067567825, -0.03173828, 0.012527466, 0.0...
http://arxiv.org/abs/2305.07969v2
GPT-Sentinel: Distinguishing Human and ChatGPT Generated Content
Yutian Chen, Hao Kang, Vivian Zhai, Liangze Li, Rita Singh, Bhiksha Raj
cs.CL
This paper presents a novel approach for detecting ChatGPT-generated vs. human-written text using language models. To this end, we first collected and released a pre-processed dataset named OpenGPTText, which consists of rephrased content generated using ChatGPT. We then designed, implemented, and trained two different...
http://arxiv.org/pdf/2305.07969v2
2023-05-17T18:21:03
2023-05-13T17:12:11
{ "title": [ 0.0024604797, -0.008605957, -0.015975952, -0.019989014, -0.0059547424, -0.009132385, -0.014511108, -0.026763916, -0.0231781, 0.02394104, -0.012435913, 0.02760315, -0.027862549, -0.019866943, -0.02027893, 0.016937256, -0.029449463, 0....
http://arxiv.org/abs/2305.07961v2
Leveraging Large Language Models in Conversational Recommender Systems
Luke Friedman, Sameer Ahuja, David Allen, Zhenning Tan, Hakim Sidahmed, Changbo Long, Jun Xie, Gabriel Schubiner, Ajay Patel, Harsh Lara, Brian Chu, Zexi Chen, Manoj Tiwari
cs.IR, cs.CL, cs.LG
A Conversational Recommender System (CRS) offers increased transparency and control to users by enabling them to engage with the system through a real-time multi-turn dialogue. Recently, Large Language Models (LLMs) have exhibited an unprecedented ability to converse naturally and incorporate world knowledge and common...
http://arxiv.org/pdf/2305.07961v2
2023-05-16T21:21:46
2023-05-13T16:40:07
{ "title": [ -0.0011072159, -0.011039734, -0.040527344, -0.060058594, 0.0026226044, -0.0018119812, -0.048309326, 0.012207031, 0.005897522, 0.036010742, -0.02255249, 0.00048184395, 0.0016393661, -0.0074310303, -0.04647827, 0.0025081635, -0.005897522, ...
http://arxiv.org/abs/2305.07928v1
AMTSS: An Adaptive Multi-Teacher Single-Student Knowledge Distillation Framework For Multilingual Language Inference
Qianglong Chen, Feng Ji, Feng-Lin Li, Guohai Xu, Ming Yan, Ji Zhang, Yin Zhang
cs.CL, cs.AI
Knowledge distillation is of key importance to launching multilingual pre-trained language models for real applications. To support cost-effective language inference in multilingual settings, we propose AMTSS, an adaptive multi-teacher single-student distillation framework, which allows distilling knowledge from multip...
http://arxiv.org/pdf/2305.07928v1
2023-05-13T14:42:30
2023-05-13T14:42:30
{ "title": [ -0.031219482, -0.023132324, -0.049743652, -0.07788086, -0.022949219, 0.00680542, -0.09106445, 0.021575928, 0.010559082, 0.015037537, -0.016082764, -0.018875122, 0.013389587, -0.019042969, -0.026519775, -0.015197754, -0.0050582886, 0....
http://arxiv.org/abs/2305.07927v1
RC3: Regularized Contrastive Cross-lingual Cross-modal Pre-training
Chulun Zhou, Yunlong Liang, Fandong Meng, Jinan Xu, Jinsong Su, Jie Zhou
cs.CL
Multilingual vision-language (V&L) pre-training has achieved remarkable progress in learning universal representations across different modalities and languages. In spite of recent success, there still remain challenges limiting further improvements of V&L pre-trained models in multilingual settings. Particularly, curr...
http://arxiv.org/pdf/2305.07927v1
2023-05-13T14:41:05
2023-05-13T14:41:05
{ "title": [ 0.0064468384, 0.017318726, -0.027236938, -0.06463623, 0.0064048767, -0.0074920654, -0.06500244, -0.028930664, 0.00034451485, 0.043304443, -0.008934021, 0.019424438, 0.020309448, -0.036621094, -0.004852295, 0.009475708, 0.0030345917, ...
http://arxiv.org/abs/2305.07922v2
CodeT5+: Open Code Large Language Models for Code Understanding and Generation
Yue Wang, Hung Le, Akhilesh Deepak Gotmare, Nghi D. Q. Bui, Junnan Li, Steven C. H. Hoi
cs.CL, cs.LG, cs.PL
Large language models (LLMs) pretrained on vast source code have achieved prominent progress in code intelligence. However, existing code LLMs have two main limitations in terms of architecture and pretraining tasks. First, they often adopt a specific architecture (encoder-only or decoder-only) or rely on a unified enc...
http://arxiv.org/pdf/2305.07922v2
2023-05-20T07:27:15
2023-05-13T14:23:07
{ "title": [ -0.039520264, -0.02772522, -0.03213501, -0.080322266, 0.000034928322, 0.018997192, -0.06964111, 0.010612488, 0.0057868958, 0.020004272, -0.008712769, 0.0062446594, 0.002588272, -0.029647827, -0.016738892, 0.005256653, -0.010971069, 0...
http://arxiv.org/abs/2305.07912v1
Pre-trained Language Model with Prompts for Temporal Knowledge Graph Completion
Wenjie Xu, Ben Liu, Miao Peng, Xu Jia, Min Peng
cs.CL, cs.AI, I.2.4; I.2.7
Temporal Knowledge graph completion (TKGC) is a crucial task that involves reasoning at known timestamps to complete the missing part of facts and has attracted more and more attention in recent years. Most existing methods focus on learning representations based on graph neural networks while inaccurately extracting i...
http://arxiv.org/pdf/2305.07912v1
2023-05-13T12:53:11
2023-05-13T12:53:11
{ "title": [ -0.010505676, -0.020263672, -0.06616211, -0.117126465, -0.006401062, -0.003955841, -0.08892822, 0.04437256, -0.0064201355, 0.035491943, -0.017852783, 0.008033752, 0.0146484375, -0.0070533752, -0.045928955, 0.0038146973, 0.0049552917, ...
http://arxiv.org/abs/2305.07895v4
On the Hidden Mystery of OCR in Large Multimodal Models
Yuliang Liu, Zhang Li, Hongliang Li, Wenwen Yu, Yang Liu, Biao Yang, Mingxin Huang, Dezhi Peng, Mingyu Liu, Mingrui Chen, Chunyuan Li, Xucheng Yin, Cheng-lin Liu, Lianwen Jin, Xiang Bai
cs.CV, cs.CL
Large models have recently played a dominant role in natural language processing and multimodal vision-language learning. It remains less explored about their efficacy in text-related visual tasks. We conducted a comprehensive study of existing publicly available multimodal models, evaluating their performance in text ...
http://arxiv.org/pdf/2305.07895v4
2023-06-19T03:36:08
2023-05-13T11:28:37
{ "title": [ -0.04763794, 0.014884949, -0.040527344, -0.016342163, -0.017913818, -0.00881958, -0.042053223, -0.033172607, -0.010551453, 0.038726807, -0.026123047, -0.013626099, 0.017745972, -0.01449585, -0.031173706, -0.011131287, -0.010231018, 0...
http://arxiv.org/abs/2305.07893v2
PESTS: Persian_English Cross Lingual Corpus for Semantic Textual Similarity
Mohammad Abdous, Poorya Piroozfar, Behrouz Minaei Bidgoli
cs.CL, cs.AI
One of the components of natural language processing that has received a lot of investigation recently is semantic textual similarity. In computational linguistics and natural language processing, assessing the semantic similarity of words, phrases, paragraphs, and texts is crucial. Calculating the degree of semantic r...
http://arxiv.org/pdf/2305.07893v2
2023-09-29T16:12:29
2023-05-13T11:02:50
{ "title": [ -0.024597168, -0.0074043274, -0.025558472, -0.05496216, 0.014724731, 0.012641907, -0.027740479, -0.013755798, 0.014373779, 0.035491943, -0.02722168, 0.02911377, 0.029403687, -0.011795044, -0.024993896, 0.026947021, -0.02357483, 0.052...
http://arxiv.org/abs/2305.07882v1
Dual Use Concerns of Generative AI and Large Language Models
Alexei Grinbaum, Laurynas Adomaitis
cs.CY, cs.CL
We suggest the implementation of the Dual Use Research of Concern (DURC) framework, originally designed for life sciences, to the domain of generative AI, with a specific focus on Large Language Models (LLMs). With its demonstrated advantages and drawbacks in biological research, we believe the DURC criteria can be eff...
http://arxiv.org/pdf/2305.07882v1
2023-05-13T10:08:57
2023-05-13T10:08:57
{ "title": [ -0.009521484, -0.013931274, -0.032409668, -0.045196533, -0.031051636, -0.018997192, -0.024291992, 0.0065231323, -0.006298065, 0.042877197, -0.037261963, 0.0016555786, 0.013916016, -0.020217896, -0.024749756, 0.008888245, -0.0015916824, ...
http://arxiv.org/abs/2305.07868v1
Bridging History with AI A Comparative Evaluation of GPT 3.5, GPT4, and GoogleBARD in Predictive Accuracy and Fact Checking
Davut Emre Tasar, Ceren Ocal Tasar
cs.CL, cs.AI
The rapid proliferation of information in the digital era underscores the importance of accurate historical representation and interpretation. While artificial intelligence has shown promise in various fields, its potential for historical fact-checking and gap-filling remains largely untapped. This study evaluates the ...
http://arxiv.org/pdf/2305.07868v1
2023-05-13T08:58:37
2023-05-13T08:58:37
{ "title": [ 0.03491211, 0.040771484, -0.026275635, -0.056549072, 0.004108429, -0.01902771, -0.024230957, -0.016021729, -0.022018433, 0.028793335, -0.05343628, 0.0050086975, 0.025115967, -0.021148682, 0.0103302, -0.029785156, -0.026977539, 0.0291...
http://arxiv.org/abs/2305.07839v1
The Geometry of Multilingual Language Models: An Equality Lens
Cheril Shah, Yashashree Chandak, Manan Suri
cs.CL
Understanding the representations of different languages in multilingual language models is essential for comprehending their cross-lingual properties, predicting their performance on downstream tasks, and identifying any biases across languages. In our study, we analyze the geometry of three multilingual language mode...
http://arxiv.org/pdf/2305.07839v1
2023-05-13T05:19:15
2023-05-13T05:19:15
{ "title": [ -0.044799805, -0.0132369995, -0.029037476, -0.054901123, -0.0063972473, 0.011993408, -0.06567383, 0.011062622, -0.0033092499, 0.054351807, 0.00048089027, -0.0044898987, 0.026855469, -0.01499176, -0.05657959, 0.018066406, -0.022445679, ...
http://arxiv.org/abs/2305.07826v1
Frequency-aware Dimension Selection for Static Word Embedding by Mixed Product Distance
Lingfeng Shen, Haiyun Jiang, Lemao Liu, Ying Chen
cs.CL
Static word embedding is still useful, particularly for context-unavailable tasks, because in the case of no context available, pre-trained language models often perform worse than static word embeddings. Although dimension is a key factor determining the quality of static word embeddings, automatic dimension selection...
http://arxiv.org/pdf/2305.07826v1
2023-05-13T02:53:37
2023-05-13T02:53:37
{ "title": [ -0.033111572, -0.0033283234, -0.043151855, -0.018508911, -0.0143585205, -0.0059509277, -0.03277588, -0.006134033, 0.005695343, 0.025024414, -0.0181427, 0.024307251, 0.008453369, -0.023361206, -0.0030384064, 0.014984131, 0.004611969, ...
http://arxiv.org/abs/2305.07824v1
A Simple and Plug-and-play Method for Unsupervised Sentence Representation Enhancement
Lingfeng Shen, Haiyun Jiang, Lemao Liu, Shuming Shi
cs.CL, cs.AI
Generating proper embedding of sentences through an unsupervised way is beneficial to semantic matching and retrieval problems in real-world scenarios. This paper presents Representation ALchemy (RepAL), an extremely simple post-processing method that enhances sentence representations. The basic idea in RepAL is to de-...
http://arxiv.org/pdf/2305.07824v1
2023-05-13T02:43:59
2023-05-13T02:43:59
{ "title": [ 0.012062073, 0.016830444, -0.027511597, -0.055603027, 0.030883789, 0.0053977966, -0.035827637, -0.02053833, 0.01360321, 0.03286743, -0.01158905, 0.00944519, 0.008003235, -0.006629944, -0.04763794, 0.0149383545, -0.022491455, 0.057464...
http://arxiv.org/abs/2305.07804v4
Improving Small Language Models on PubMedQA via Generative Data Augmentation
Zhen Guo, Peiqi Wang, Yanwei Wang, Shangdi Yu
cs.CL, cs.LG
Large Language Models (LLMs) have made remarkable advancements in the field of natural language processing. However, their increasing size poses challenges in terms of computational cost. On the other hand, Small Language Models (SLMs) are known for their efficiency, but they often struggle with limited capacity and tr...
http://arxiv.org/pdf/2305.07804v4
2023-08-01T20:27:56
2023-05-12T23:49:23
{ "title": [ -0.028701782, -0.011734009, -0.040740967, -0.050476074, -0.018814087, 0.003232956, -0.06451416, -0.0012741089, 0.013305664, 0.029647827, 0.0047683716, -0.009353638, -0.009963989, -0.038482666, -0.0541687, 0.009315491, -0.0029335022, ...
http://arxiv.org/abs/2305.07797v2
ACCENT: An Automatic Event Commonsense Evaluation Metric for Open-Domain Dialogue Systems
Sarik Ghazarian, Yijia Shao, Rujun Han, Aram Galstyan, Nanyun Peng
cs.CL
Commonsense reasoning is omnipresent in human communications and thus is an important feature for open-domain dialogue systems. However, evaluating commonsense in dialogue systems is still an open challenge. We take the first step by focusing on event commonsense that considers events and their relations, and is crucia...
http://arxiv.org/pdf/2305.07797v2
2023-11-03T05:06:06
2023-05-12T23:11:48
{ "title": [ 0.010360718, -0.0062294006, -0.01651001, -0.055541992, 0.015090942, -0.009162903, -0.01676941, -0.008636475, 0.02420044, 0.018432617, 0.004016876, 0.004600525, 0.00062942505, -0.033447266, -0.040130615, 0.009864807, -0.030426025, 0.0...
http://arxiv.org/abs/2305.07795v2
Constructing Holistic Measures for Social Biases in Masked Language Models
Yang Liu, Yuexian Hou
cs.CL
Masked Language Models (MLMs) have been successful in many natural language processing tasks. However, real-world stereotype biases are likely to be reflected in MLMs due to their learning from large text corpora. Most of the evaluation metrics proposed in the past adopt different masking strategies, designed with the ...
http://arxiv.org/pdf/2305.07795v2
2023-09-01T13:44:14
2023-05-12T23:09:06
{ "title": [ -0.038757324, 0.002954483, -0.06732178, -0.016052246, -0.007408142, -0.027145386, -0.04220581, -0.013656616, 0.0052337646, 0.019638062, -0.025558472, 0.008735657, 0.028182983, -0.022399902, -0.013549805, 0.0058250427, -0.034606934, 0...
http://arxiv.org/abs/2305.07789v1
Answering Complex Questions over Text by Hybrid Question Parsing and Execution
Ye Liu, Semih Yavuz, Rui Meng, Dragomir Radev, Caiming Xiong, Yingbo Zhou
cs.CL, cs.AI
The dominant paradigm of textual question answering systems is based on end-to-end neural networks, which excels at answering natural language questions but falls short on complex ones. This stands in contrast to the broad adaptation of semantic parsing approaches over structured data sources (e.g., relational database...
http://arxiv.org/pdf/2305.07789v1
2023-05-12T22:37:06
2023-05-12T22:37:06
{ "title": [ -0.01449585, -0.0058937073, -0.021102905, -0.06573486, 0.0050849915, 0.009376526, -0.030807495, 0.0042648315, -0.0011720657, 0.019165039, -0.030883789, 0.010414124, -0.014091492, -0.005039215, -0.055511475, -0.01499939, -0.0022697449, ...