Datasets:

havens2

/

naacl2022

like 0

[ "3.2", "Model", "Overview" ]

[ 0, 0, 0 ]

[ "We", "also", "experimented", "with", "a", "sentiment", "classification", "dataset", "which", "provides", "sentiment", "labels", "consisting", "of", "positive,", "negative", "and", "neutral." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "Emotions", "are", "labeled", "as", "one", "of", "the", "predefined", "classes", "depending", "on", "the", "dataset,", "and", "the", "emotions", "we", "experimented", "with", "are", "either", "6", "or", "7." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "The", "ERC", "is", "a", "task", "of", "predicting", "the", "emotion", "et", "of", "ut,", "the", "utterance", "of", "the", "t-th", "turn,", "given", "the", "previous", "utterances", "ht", "=", "." ]

[ 0, 11, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "While", "pui", "and", "puj", "(i", "=", "j)", "can", "be", "the", "same", "speaker,", "the", "minimum", "number", "of", "the", "unique", "conversation", "participants", "should", "be", "2", "or", "more." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "ui", "is", "the", "utterance", "which", "the", "speaker", "pui", "uttered,", "where", "pui", "is", "one", "of", "the", "conversation", "participants." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "In", "a", "conversation,", "M", "sequential", "utterances", "are", "given", "as", "[(u1,", "pu1),", "(u2,", "pu2),", "...,", "(uM", ",", "puM", ")]." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "3.1", "Problem", "Statement" ]

[ 0, 0, 0 ]

[ "3", "Approach" ]

[ 0, 0 ]

[ "EDA", "(Bothe", "et", "al.,", "2020)", "expands", "the", "multi-modal", "emotion", "datasets", "by", "extracting", "dialog", "acts", "from", "MELD", "and", "IEMOCAP", "and", "finds", "out", "that", "there", "is", "a", "correlation", "between", "dialogue", "acts", "and", "emotion", "labels." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 13, 0, 13, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "Liu", "et", "al.", "(2020)", "reports", "that", "the", "lack", "of", "external", "knowledge", "makes", "it", "difficult", "to", "classify", "implicit", "emotions", "from", "the", "conversation", "history." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "Zhou", "et", "al.", "(2018);", "Zhang", "et", "al.", "(2018a)", "shows", "that", "commonsense", "knowledge", "is", "important", "for", "understanding", "conversations", "and", "generating", "appropriate", "responses." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "dyadic-party", "dialogues,", "therefore", "have", "more", "conditions", "to", "consider", "and", "result", "in", "poor", "performance." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "In", "general,", "the", "multi-party", "conversations", "have", "higher", "speaker", "dependency", "than", "the" ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "However,", "as", "the", "multi-party", "conversation", "datasets", "including", "MELD", "and", "EmoryNLP", "have", "become", "available,", "a", "lot", "of", "recent", "research", "is", "being", "conducted", "on", "multi-party", "dialogues", "such", "as", "Zhang", "et", "al." ]

[ 0, 0, 0, 0, 0, 0, 0, 13, 0, 13, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ ";", "Zadeh", "et", "al.", "(2017);", "Majumder", "et", "al.", "(2019),", "most", "works", "focused", "on", "dyadic-party", "conversation." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "In", "this", "work,", "we", "design", "and", "introduce", "a", "text-based", "emotion", "recognition", "system", "using", "neural", "networks." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 11, 12, 0, 0, 0, 0 ]

[ "Multimodal", "information", "is", "not", "always", "given", "in", "most", "social", "media,", "especially", "in", "chatbot", "systems", "where", "they", "are", "mainly", "composed", "of", "text-based", "systems." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "MELD", "and", "ICON", "(Hazarika", "et", "al.,", "2018a)", "show", "that", "the", "more", "multi-modal", "information", "is", "used,", "the", "better", "the", "performance", "and", "the", "text", "information", "plays", "the", "most", "important", "role." ]

[ 13, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "Datcu", "and", "Rothkrantz", "(2014)", "uses", "speech", "and", "visual", "information", "to", "recognize", "emotions,", "and", "(Alm", "et", "al.,", "2005)", "attempts", "to", "recognize", "emotions", "based", "on", "text", "information." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "The", "multi-modal", "data", "such", "as", "MELD", "and", "IEMOCAP", "are", "some", "of", "the", "available", "standard", "datasets", "for", "emotion", "recognition", "and", "they", "are", "composed", "of", "text,", "speech", "and", "vision-based", "data." ]

[ 0, 0, 0, 0, 0, 13, 0, 13, 0, 0, 0, 0, 0, 0, 0, 0, 11, 12, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "In", "addition,", "Ekman", "explains", "that", "a", "multimodal", "view", "is", "important", "for", "multiple", "emotions", "recognition." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "Ekman", "(Ekman,", "1992)", "constructs", "taxonomy", "of", "six", "common", "emotions", "(Joy,", "Sadness,", "Fear,", "Anger,", "Surprise,", "and", "Disgust)", "from", "human", "facial", "expressions." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "ToDKAT", "(Zhu", "et", "al.,", "2021)", "improves", "performance", "by", "combining", "commonsense", "knowledge", "using", "COMET", "and", "topic", "discovery", "using", "VHRED", "(Serban", "et", "al.,", "2017)", "to", "the", "model." ]

[ 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "ATOMIC", "has", "9", "sentence", "relation", "types", "with", "inferential", "if-then", "commonsense", "knowledge", "expressed", "in", "text." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "Commonsense", "knowledge", "feature", "is", "extracted", "and", "leveraged", "with", "COMET", "(Bosselut", "et", "al.,", "2019)", "trained", "with", "ATOMIC", "(The", "Atlas", "of", "Machine", "Commonsense)", "(Sap", "et", "al.,", "2019)." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "COSMIC", "(Ghosal", "et", "al.,", "2020)", "and", "Psychological", "(Li", "et", "al.,", "2021)", "improve", "the", "performance", "of", "emotion", "recognition", "by", "extracting", "commonsense", "knowledge", "of", "the", "previous", "utterances." ]

[ 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 11, 12, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "NRC_VAD", "Lexicon", "has", "human", "ratings", "of", "valence,", "arousal,", "and", "dominance", "for", "more", "than", "20,000", "English", "words." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "ConceptNet", "is", "a", "knowledge", "graph", "that", "connects", "words", "and", "phrases", "in", "natural", "language", "using", "labeled", "edges." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "KET", "(Zhong", "et", "al.,", "2019)", "is", "used", "as", "external", "knowledge", "based", "on", "ConceptNet", "(Speer", "et", "al.,", "2017)", "and", "emotion", "lex" ]

[ 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "Many", "recent", "studies", "use", "external", "knowledge", "to", "improve", "the", "ERC", "performance." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 11, 0 ]

[ "2", "Related", "Work" ]

[ 0, 0, 0 ]

[ "Further", "experiments", "also", "show", "that", "our", "approach", "can", "be", "used", "in", "other", "languages", "and", "show", "the", "performance", "of", "CoMPM", "when", "the", "number", "of", "data", "is", "limited." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0 ]

[ "We", "perform", "an", "ablation", "study", "on", "each", "module", "to", "show", "that", "the", "proposed", "approach", "is", "effective." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "CoMPM", "achieves", "the", "first", "or", "second", "performance", "according", "to", "the", "evaluation", "metric", "compared", "to", "all", "previous", "systems." ]

[ 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "Multi-party", "datasets", "are", "MELD", "(Poria", "et", "al.,", "2019)", "and", "EmoryNLP", "(Zahiri", "and", "Choi,", "2018),", "and", "dyadic", "datasets", "are", "IEMOCAP", "(Busso", "et", "al.,", "2008)", "and", "DailyDialog", "(Li", "et", "al.,", "2017)." ]

[ 0, 0, 0, 13, 0, 0, 0, 0, 0, 13, 0, 0, 0, 0, 0, 0, 0, 0, 13, 0, 0, 0, 0, 0, 13, 0, 0, 0, 0 ]

[ "We", "experiment", "on", "4", "different", "English", "ERC", "datasets." ]

[ 0, 0, 0, 0, 0, 0, 11, 0 ]

[ "We", "use", "the", "PM", "to", "help", "predict", "the", "emotion", "of", "the", "speaker", "by", "taking", "into", "account", "the", "speaker’s", "linguistic", "preferences", "and", "characteristics." ]

[ 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "We", "use", "the", "output", "of", "the", "pre-trained", "language", "model", "as", "the", "memory", "embedding", "where", "the", "utterances", "are", "passed", "into", "the", "language", "model." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "The", "second", "is", "a", "pre-trained", "memory", "module", "(PM)", "that", "extracts", "memory", "from", "utterances." ]

[ 0, 0, 0, 0, 1, 2, 2, 1, 0, 0, 0, 0, 0 ]

[ "(1)", "The", "first", "is", "a", "context", "embedding", "module", "(CoM)", "that", "reflects", "all", "previous", "utterances", "as", "context." ]

[ 0, 0, 0, 0, 0, 1, 2, 2, 1, 0, 0, 0, 0, 0, 0, 0 ]

[ "CoM", "is", "an", "auto-regressive", "model", "that", "predicts", "the", "current", "emotion", "through", "attention", "between", "the", "previous", "utterances", "of", "the", "conversation", "and", "the", "current", "utterance." ]

[ 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "CoMPM,", "introduced", "in", "this", "paper,", "is", "composed", "of", "two", "modules", "that", "take", "into", "account", "previous", "utterances", "in", "dialogue." ]

[ 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "Based", "on", "these", "studies,", "we", "eliminate", "the", "dependence", "on", "structured", "external", "data", "used", "in", "cutting-edge", "systems", "and", "use", "a", "pre-trained", "language", "model", "as", "a", "feature", "extractor", "of", "knowledge." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "Petroni", "et", "al.", "(2019)", "introduces", "that", "these", "language", "models", "can", "be", "used", "as", "knowledge", "bases", "and", "have", "many", "advantages", "over", "the", "structured", "knowledge", "bases." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "Since", "pre-trained", "language", "models", "are", "trained", "by", "unsupervised", "learning,", "these", "models", "are", "relatively", "usable", "approaches", "regardless", "of", "language", "types." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "In", "recent", "NLP", "studies,", "due", "to", "the", "effectiveness", "of", "the", "pre-trained", "language", "model,", "it", "has", "already", "been", "developed", "in", "many", "countries." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "In", "order", "to", "utilize", "the", "previous", "methods", "in", "languages", "of", "other", "countries,", "it", "is", "expensive", "and", "difficult", "to", "utilize", "because", "external", "knowledge", "data", "must", "be", "newly", "constructed." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "Many", "recent", "studies", "use", "external", "knowledge", "to", "improve", "the", "ERC", "performance." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 11, 0 ]

[ "AGHMN", "(Jiao", "et", "al.,", "2020)", "considers", "the", "previous", "utterances", "through", "memory", "summarizing", "using", "GRU", "with", "attention." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "Representatively,", "DialogueRNN", "(Majumder", "et", "al.,", "2019)", "recognizes", "the", "present", "emotion", "by", "tracking", "context", "from", "the", "previous", "utterances", "and", "the", "speaker’s", "emotion." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "Therefore,", "recent", "studies", "are", "attempting", "to", "recognize", "emotions", "while", "taking", "into", "account", "the", "previous", "utterances." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "Like", "the", "previous", "studies", "(Ghosal", "et", "al.,", "2020),", "we", "show", "that", "the", "utterance-level", "emotion", "recognition,", "which", "does", "not", "consider", "the", "previous", "utterance,", "have", "limitations", "and", "experiments", "result", "in", "poor", "performances." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "If", "the", "system", "does", "not", "take", "into", "account", "previous", "utterances,", "it", "is", "difficult", "to", "properly", "recognize", "emotions." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "The", "emotion", "of", "speaker", "B’s", "utterance", "(\"How’d", "you", "get", "to", "that?\")", "is", "angry." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "Figure", "1", "is", "an", "example", "of", "a", "conversation", "in", "which", "two", "speakers", "are", "angry", "at", "each", "other." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "As", "these", "interactive", "machines", "increase,", "the", "ERC", "module", "plays", "an", "increasingly", "important", "role." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "In", "addition,", "emotion", "recognition", "can", "be", "effectively", "used", "for", "opinion", "mining,", "recommender", "systems,", "and", "healthcare", "systems", "where", "it", "can", "improve", "the", "service", "qualities", "by", "providing", "personalized", "results." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "As", "the", "number", "of", "applications", "such", "as", "interactive", "chatbots", "or", "social", "media", "that", "are", "used", "by", "many", "users", "has", "recently", "increased", "dramatically,", "Emotion", "Recognition", "in", "Conversation", "(ERC)", "plays", "a", "more", "important", "role", "in", "natural", "language", "processing,", "and", "as", "a", "proof,", "a", "lot", "of", "research", "(Poria", "et", "al.,", "2019;", "Zhang", "et", "al.,", "2019;", "Ghosal", "et", "al.,", "2020;", "Jiao", "et", "al.,", "2020)", "has", "been", "conducted", "on", "the", "task." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 11, 12, 12, 12, 11, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "Introduction" ]

[ 0 ]

[ "Our", "code", "is", "available", "on", "github", "1." ]

[ 0, 0, 0, 0, 0, 0, 0 ]

[ "In", "addition,", "our", "method", "shows", "that", "it", "can", "be", "extended", "to", "other", "languages", "because", "structured", "knowledge", "is", "not", "required,", "unlike", "previous", "methods." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "CoMPM", "achieves", "the", "first", "or", "second", "performance", "on", "all", "data", "and", "is", "state-of-the-art", "among", "systems", "that", "do", "not", "leverage", "structured", "data." ]

[ 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "We", "introduce", "CoMPM,", "which", "combines", "the", "speaker’s", "pre-trained", "memory", "with", "the", "context", "model,", "and", "find", "that", "the", "pre-trained", "memory", "significantly", "improves", "the", "performance", "of", "the", "context", "model." ]

[ 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "Therefore,", "we", "extract", "the", "pre-trained", "memory", "using", "the", "pre-trained", "language", "model", "as", "an", "extractor", "of", "external", "knowledge." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "However,", "structured", "data", "is", "difficult", "to", "access", "in", "non-English", "languages,", "making", "it", "difficult", "to", "extend", "to", "other", "languages." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "Many", "recent", "approaches", "show", "performance", "improvement", "by", "combining", "knowledge", "into", "modules", "learned", "from", "external", "structured", "data." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "Since", "emotion", "recognition", "in", "conversation", "is", "inaccurate", "if", "the", "previous", "utterances", "are", "not", "taken", "into", "account,", "many", "studies", "reflect", "the", "dialogue", "context", "to", "improve", "the", "performances." ]

[ 0, 11, 12, 12, 12, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "If", "the", "machine-generated", "sentences", "reflect", "emotion,", "more", "human-like", "sympathetic", "conversations", "are", "possible." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "As", "the", "use", "of", "interactive", "machines", "grow,", "the", "task", "of", "Emotion", "Recognition", "in", "Conversation", "(ERC)", "became", "more", "important." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 11, 12, 12, 12, 11, 0, 0, 0 ]

[ "Abstract" ]

[ 0 ]

[ "CoMPM:", "Context", "Modeling", "with", "Speaker’s", "Pre-trained", "Memory", "Tracking", "for", "Emotion", "Recognition", "in", "Conversation" ]

[ 1, 0, 0, 0, 0, 0, 0, 0, 0, 11, 12, 12, 12 ]

[ "Figure", "4b", "demonstrates", "a", "termination", "of", "stock", "sale" ]

[ 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "Herein,", "we", "visualize", "the", "heatmaps", "in", "some", "cases", "to", "investigate", "how", "to", "assign", "attention", "weights", "to", "text." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "A.4", "Case", "Study:", "Visualization", "of", "Attention" ]

[ 0, 0, 0, 0, 0, 0 ]

[ "We", "summarize", "the", "hyperparameters", "in", "Table", "5." ]

[ 0, 0, 0, 0, 0, 0, 0 ]

[ "In", "addition,", "the", "coefficient", "λ", "of", "the", "regularization", "term", "was", "set", "as", "0.0001." ]

[ 0, 0, 0, 3, 3, 0, 0, 0, 0, 0, 0, 0, 5 ]

[ "We", "used", "the", "Adam", "optimizer", "with", "learning", "rates", "of", "0.1", "and", "0.001", "in", "the", "inner", "and", "outer", "updates,", "that", "is,", "α", "and", "β", "in", ",", "respectively." ]

[ 0, 0, 0, 0, 0, 0, 3, 4, 0, 5, 0, 5, 0, 0, 0, 0, 0, 0, 0, 0, 3, 0, 3, 0, 0, 0 ]

[ "After", "that,", "the", "performance", "of", "the", "adapted", "task", "model", "is", "obtained", "using", "queries." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "In", "validation", "and", "test", "process,", "we", "val", "and", "sampled", "30", "tasks", "with", "15", "queries", "from", "test,", "and", "only", "performed", "task", "adaptation", "using", "S", "K-shots." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "During", "the", "meta-training", "process,", "we", "sampled", "four", "tasks", "with", "15", "queries", "tr,", "and", "it", "leads", "to", "performing", "task", "adaptation", "from", "four", "times", "per", "each", "meta-optimization", "update", "and", "early", "stop", "when", "the", "validation", "loss", "fails", "to", "improve", "for", "20", "steps." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "In", "our", "work,", "we", "train", "all", "experiments", "on", "a", "single", "NVIDIA", "A100", "32G", "GPU." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "A.3", "Training", "Details", "and", "Hyperparameter" ]

[ 0, 0, 0, 0, 0 ]

[ "Table", "3", "summarizes", "the", "above", "model", "parameters." ]

[ 0, 0, 0, 0, 0, 0, 0 ]

[ "We", "set", "the", "size", "to", "150", "for", "the", "meta-attention-aspects", "dictionary,", "and", "importantly", "fixed", "the", "number", "of", "top-k", "attention", "aspects", "to", "20." ]

[ 0, 0, 0, 0, 0, 5, 0, 0, 0, 0, 0, 0, 0, 0, 3, 4, 4, 4, 4, 0, 5 ]

[ "For", "each", "task", "classifier,", "that", "is,", "fθ′τi", ",", "it", "is", "designed", "as", "single-layer", "fully", "connected", "neural", "network." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "The", "gating", "network,", "Wg", "is", "linear", "transformation", "and", "its", "size", "is", "the", "number", "of", "meta-attention-aspects." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 3, 4, 4 ]

[ "As", "shown", "in", "these", "cases,", "LEA", "properly", "captures", "important", "words", "under", "a", "certain", "topic", "and", "assigns", "attention", "weights", "to", "a", "given", "text." ]

[ 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "Figure", "4d", "talks", "about", "the", "authority", "of", "platinum", "and", "gold", "coins", "under", "the", "Economics", "topic,", "and", "the", "words", "“US,”", "“Mint,”", "“authority,”", "“gold,”", "“platinum,”", "and", "“coin”", "are", "hence", "highlighted." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "Figure", "4c", "shows", "that", "the", "Turkish", "market", "was", "closed", "related", "to", "the", "Market", "topic,", "and", "its", "relevant", "words", "such", "as", "“Turkish,”", "“markets,”", "and", "“closed”", "are", "highly", "attended", "as", "expected." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "These", "were", "extracted", "under", "the", "Corporate", "and", "Industrial", "topic", "in", "the", "RCV-1", "dataset", "and", "some", "seminal", "words", "such", "as", "“agreement”,", "“contractual”", "and", "“receivership”", "are", "highlighted", "to", "appear", "in", "the", "topic." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 13, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "The", "texts", "in", "(a)", "and", "(b)", "are", "different", "samples", "under", "a", "topic", "Corporate", "and", "Industrial." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "The", "relation", "network,", "fθr", "is", "composed", "of", "two-layers", "neural", "network", "with", "ReLU", "activation", "and", "input", "size", "is", "two", "times", "of", "encoder", "outputs", "and", "the", "size", "of", "output", "is", "the", "number", "of", "meta-attention-aspects,", "i.e.,", "150." ]

[ 0, 0, 0, 0, 0, 0, 0, 5, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 3, 4, 4, 0, 0, 3, 4, 4, 0, 5 ]

[ "In", "the", "encoding", "process", "of", "our", "experiments,", "the", "768-dimensional", "[CLS]", "vector,", "which", "is", "of", "the", "same", "size", "of", "the", "output", "of", "the", "pre-trained", "BERTbase-uncased,", "is", "linearly", "transformed", "through", "fθe", "into", "a", "300-dimensional", "vector." ]

[ 0, 0, 0, 0, 0, 0, 0, 0, 5, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 5, 0 ]

[ "We", "share", "the", "breakdown", "of", "LEA’s", "implementation." ]

[ 0, 0, 0, 0, 0, 1, 0 ]

[ "A.2", "Implementation", "Details" ]

[ 0, 0, 0 ]

[ "We", "summarize", "the", "above", "information", "in", "Table", "4." ]

[ 0, 0, 0, 0, 0, 0, 0, 0 ]

[ "For", "the", "20", "Newsgroup", "dataset,", "we", "randomly", "selected", "20", "topic", "classes,", "and", "the", "metatraining", "set,", "meta-validation", "set,", "and", "meta-test", "set", "contained", "10,", "5,", "and", "5", "disjoint", "classes,", "respectively." ]

[ 0, 0, 13, 14, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 5, 6, 6, 6, 0, 0, 0 ]

[ "As", "a", "result,", "Amazon", "product", "data", "is", "divided", "into", "15,", "and", "9", "disjoint", "classes", "for", "meta-training", "and", "test", "sets", "and", "meta-validation", "set", "is", "not", "used", "in", "Amazon", "product", "data." ]

[ 0, 0, 0, 13, 14, 14, 0, 0, 0, 5, 6, 6, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 13, 14, 14 ]

[ "In", "Amazon", "product", "data,", "we", "split", "the", "data", "using", "rules", "in", "(Bao", "et", "al.,", "2019),", "and", "its", "training", "and" ]

[ 0, 13, 14, 14, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ]