table_result/2407.00119v2_output.json

[
    {
        "path": "table_paper/2407.00119v2.json",
        "table_id": "1",
        "section": "4.3",
        "all_context": [
            "To verify the superiority of the ELR-GNN method proposed in this paper, we report the experimental results of ELR-GNN and other comparative methods on the IEMOCAP and MELD data sets.",
            "Experimental results are presented in Tables 1 and 2 .",
            "IEMOCAP: As shown in Table 1 , the multi-modal emotion recognition method proposed in this paper achie-ved the best emotion recognition effect on the IEMOCAP data set, with an average accuracy of 70.6% and an average F1 value of 70.9%.",
            "ELR-GCN proposes an effective modeling method of long-distance context latent dependencies for multi-modal emotion recognition.",
            "In addition, ELR-GCN also combines early and adaptive late fusion methods to achieve the capture of fine-grained emotional features.",
            "Among other comparison methods, the emotion recognition effect of DER-GCN is slightly lower than that of ELR-GNN, with an average accuracy of 69.7% and an average F1 value of 69.4%.",
            "Although DER-GCN comprehensively considers event relationships and dialogue relationships between speakers to enhance the model s emotional understanding, it ignores latent context dependencies.",
            "The emotion recognition effect of LR-GCN is lower than ELR-GNN and DER-GCN, with an average accuracy of 68.5% and an average F1 value of 68.3%.",
            "Although LR-GCN considers latent dependencies between contexts, due to the high computational complexity of GCN, LR-GCN can only capture local latent dependencies.",
            "The emotion recognition effects of other comparison methods are lower than ELR-GNN.",
            "Likewise, none of them take into account potential dependencies on context.",
            "Overall, the accuracy of ELR-GNN on the happy emotion analogy is much higher than that of other comparison algorithms, while the accuracy of other emotion categories is also relatively close to that of other comparison algorithms.",
            "In addition, the F1 value of ELR-GNN on the happy and excited emotional analogies is much higher than that of other comparison algorithms.",
            "At the same time, the F1 value of ELR-GNN on other emotional categories is also relatively close to other comparison algorithms.",
            "The experimental results prove the superiority of the ELR-GNN method proposed in this paper.",
            "MELD: As shown in Table 2 , The ELR-GNN method proposed in this article has the best emotion recognition effect on the MELD data set, with an average accuracy of 68.7% and an average F1 value of 69.9%.",
            "The emotion recognition effect of DER-GCN is second, with an average accuracy of 69.7% and an average F1 value of 69.4%.",
            "The emotion recognition effect of LR-GCN is lower than that of ELR-GNN and DER-GCN, with an average accuracy of 68.5% and an average F1 value of 68.3%.",
            "The emotion recognition effects of other comparison methods are relatively poor, and the average accuracy and F1 value are lower than ELR-GNN.",
            "The performance improvement may be attributed to ELR-GNN s ability to capture long-distance contextual latent dependencies and fine-grained fusion of dialogue relationships between speakers, contextual latent dependencies and contextual semantic information.",
            "Overall, the accuracy of ELR-GNN on the neutral, fear, sadness, joy, and disgust emotion analogy is much higher than that of other comparison algorithms, while the accuracy of other emotion categories is also relatively close to that of other comparison algorithms.",
            "In addition, the F1 value of ELR-GNN on the neutral, fear, sadness, joy, and anger emotional analogies is much higher than that of other comparison algorithms.",
            "At the same time, the F1 value of ELR-GNN on other emotional categories is also relatively close to other comparison algorithms.",
            "In addition, we find that ELR-GNN has better emotion recognition effects on the minority emotions fear and disgust, with relatively high accuracy and F1 value.",
            "The experimental results prove the superiority of the ELR-GNN method proposed in this paper.",
            "In addition, to intuitively illustrate that the running time of the ELR-GNN method proposed in this paper is better than other comparative methods, we statistics in Table 3 the running time of other comparative methods of the ELR-GNN method on the IEMOCAP and MELD data sets.",
            "As shown in Table 3 , the running time of the ELR-GNN method proposed in this paper on the IEMOCAP and MELD data sets is 41s and 91s respectively, which is significantly better than other comparison methods.",
            "The running times of DialogueGCN are 58s and 127s respectively, which are lower than LR-GCN and DER-GCN, but the emotion recognition effect is relatively poor.",
            "The running times of LR-GCN are 87s and 142s respectively.",
            "The running times of DER-GCN are 125s and 189s respectively.",
            "The experimental results prove the efficiency and effectiveness of the ELR-GNN method proposed in this paper.",
            ""
        ],
        "target_context_ids": [
            2,
            3,
            4,
            5,
            6,
            7,
            8,
            9,
            10,
            11,
            12,
            13,
            14
        ],
        "selected_paragraphs": [
            "[paragraph id = 2] IEMOCAP: As shown in Table 1 , the multi-modal emotion recognition method proposed in this paper achie-ved the best emotion recognition effect on the IEMOCAP data set, with an average accuracy of 70.6% and an average F1 value of 70.9%.",
            "[paragraph id = 3] ELR-GCN proposes an effective modeling method of long-distance context latent dependencies for multi-modal emotion recognition.",
            "[paragraph id = 4] In addition, ELR-GCN also combines early and adaptive late fusion methods to achieve the capture of fine-grained emotional features.",
            "[paragraph id = 5] Among other comparison methods, the emotion recognition effect of DER-GCN is slightly lower than that of ELR-GNN, with an average accuracy of 69.7% and an average F1 value of 69.4%.",
            "[paragraph id = 6] Although DER-GCN comprehensively considers event relationships and dialogue relationships between speakers to enhance the model s emotional understanding, it ignores latent context dependencies.",
            "[paragraph id = 7] The emotion recognition effect of LR-GCN is lower than ELR-GNN and DER-GCN, with an average accuracy of 68.5% and an average F1 value of 68.3%.",
            "[paragraph id = 8] Although LR-GCN considers latent dependencies between contexts, due to the high computational complexity of GCN, LR-GCN can only capture local latent dependencies.",
            "[paragraph id = 9] The emotion recognition effects of other comparison methods are lower than ELR-GNN.",
            "[paragraph id = 10] Likewise, none of them take into account potential dependencies on context.",
            "[paragraph id = 11] Overall, the accuracy of ELR-GNN on the happy emotion analogy is much higher than that of other comparison algorithms, while the accuracy of other emotion categories is also relatively close to that of other comparison algorithms.",
            "[paragraph id = 12] In addition, the F1 value of ELR-GNN on the happy and excited emotional analogies is much higher than that of other comparison algorithms.",
            "[paragraph id = 13] At the same time, the F1 value of ELR-GNN on other emotional categories is also relatively close to other comparison algorithms.",
            "[paragraph id = 14] The experimental results prove the superiority of the ELR-GNN method proposed in this paper."
        ],
        "table_html": "<figure class=\"ltx_table\" id=\"S4.T1\">\n<figcaption class=\"ltx_caption\"><span class=\"ltx_tag ltx_tag_table\">Table 1: </span>Comparison with other baseline models on the IEMOCAP dataset.</figcaption>\n<table class=\"ltx_tabular ltx_guessed_headers ltx_align_middle\" id=\"S4.T1.1\">\n<thead class=\"ltx_thead\">\n<tr class=\"ltx_tr\" id=\"S4.T1.1.1.1\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_column ltx_th_row ltx_border_r ltx_border_t\" id=\"S4.T1.1.1.1.1\" rowspan=\"3\" style=\"padding:2.5pt 8.7pt;\"><span class=\"ltx_text\" id=\"S4.T1.1.1.1.1.1\">Methods</span></th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" colspan=\"7\" id=\"S4.T1.1.1.1.2\" style=\"padding:2.5pt 8.7pt;\">IEMOCAP</th>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T1.1.2.2\">\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T1.1.2.2.1\" style=\"padding:2.5pt 8.7pt;\">Happy</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T1.1.2.2.2\" style=\"padding:2.5pt 8.7pt;\">Sad</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T1.1.2.2.3\" style=\"padding:2.5pt 8.7pt;\">Neutral</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T1.1.2.2.4\" style=\"padding:2.5pt 8.7pt;\">Angry</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T1.1.2.2.5\" style=\"padding:2.5pt 8.7pt;\">Excited</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T1.1.2.2.6\" style=\"padding:2.5pt 8.7pt;\">Frustrated</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T1.1.2.2.7\" style=\"padding:2.5pt 8.7pt;\">Average(w)</th>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T1.1.3.3\">\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T1.1.3.3.1\" style=\"padding:2.5pt 8.7pt;\">Acc. F1</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T1.1.3.3.2\" style=\"padding:2.5pt 8.7pt;\">Acc. F1</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T1.1.3.3.3\" style=\"padding:2.5pt 8.7pt;\">Acc. F1</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T1.1.3.3.4\" style=\"padding:2.5pt 8.7pt;\">Acc. F1</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T1.1.3.3.5\" style=\"padding:2.5pt 8.7pt;\">Acc. F1</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T1.1.3.3.6\" style=\"padding:2.5pt 8.7pt;\">Acc. F1</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T1.1.3.3.7\" style=\"padding:2.5pt 8.7pt;\">Acc. F1</th>\n</tr>\n</thead>\n<tbody class=\"ltx_tbody\">\n<tr class=\"ltx_tr\" id=\"S4.T1.1.4.1\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_r ltx_border_t\" id=\"S4.T1.1.4.1.1\" style=\"padding:2.5pt 8.7pt;\">TextCNN</th>\n<td class=\"ltx_td ltx_align_center ltx_border_t\" id=\"S4.T1.1.4.1.2\" style=\"padding:2.5pt 8.7pt;\">27.7 29..8</td>\n<td class=\"ltx_td ltx_align_center ltx_border_t\" id=\"S4.T1.1.4.1.3\" style=\"padding:2.5pt 8.7pt;\">57.1 53.8</td>\n<td class=\"ltx_td ltx_align_center ltx_border_t\" id=\"S4.T1.1.4.1.4\" style=\"padding:2.5pt 8.7pt;\">34.3 40.1</td>\n<td class=\"ltx_td ltx_align_center ltx_border_t\" id=\"S4.T1.1.4.1.5\" style=\"padding:2.5pt 8.7pt;\">61.1 52.4</td>\n<td class=\"ltx_td ltx_align_center ltx_border_t\" id=\"S4.T1.1.4.1.6\" style=\"padding:2.5pt 8.7pt;\">46.1 50.0</td>\n<td class=\"ltx_td ltx_align_center ltx_border_t\" id=\"S4.T1.1.4.1.7\" style=\"padding:2.5pt 8.7pt;\">62.9 55.7</td>\n<td class=\"ltx_td ltx_align_center ltx_border_t\" id=\"S4.T1.1.4.1.8\" style=\"padding:2.5pt 8.7pt;\">48.9 48.1</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T1.1.5.2\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_r\" id=\"S4.T1.1.5.2.1\" style=\"padding:2.5pt 8.7pt;\">bc-LSTM</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.5.2.2\" style=\"padding:2.5pt 8.7pt;\">29.1 34.4</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.5.2.3\" style=\"padding:2.5pt 8.7pt;\">57.1 60.8</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.5.2.4\" style=\"padding:2.5pt 8.7pt;\">54.1 51.8</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.5.2.5\" style=\"padding:2.5pt 8.7pt;\">57.0 56.7</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.5.2.6\" style=\"padding:2.5pt 8.7pt;\">51.1 57.9</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.5.2.7\" style=\"padding:2.5pt 8.7pt;\">67.1 58.9</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.5.2.8\" style=\"padding:2.5pt 8.7pt;\">55.2 54.9</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T1.1.6.3\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_r\" id=\"S4.T1.1.6.3.1\" style=\"padding:2.5pt 8.7pt;\">MFN</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.6.3.2\" style=\"padding:2.5pt 8.7pt;\">24.0 34.1</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.6.3.3\" style=\"padding:2.5pt 8.7pt;\">65.6 70.5</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.6.3.4\" style=\"padding:2.5pt 8.7pt;\">55.5 52.1</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.6.3.5\" style=\"padding:2.5pt 8.7pt;\">72.3 66.8</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.6.3.6\" style=\"padding:2.5pt 8.7pt;\">64.3 62.1</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.6.3.7\" style=\"padding:2.5pt 8.7pt;\">67.9 62.5</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.6.3.8\" style=\"padding:2.5pt 8.7pt;\">60.1 59.9</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T1.1.7.4\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_r\" id=\"S4.T1.1.7.4.1\" style=\"padding:2.5pt 8.7pt;\">CMN</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.7.4.2\" style=\"padding:2.5pt 8.7pt;\">25.0 30.3</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.7.4.3\" style=\"padding:2.5pt 8.7pt;\">55.9 62.4</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.7.4.4\" style=\"padding:2.5pt 8.7pt;\">52.8 52.3</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.7.4.5\" style=\"padding:2.5pt 8.7pt;\">61.7 59.8</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.7.4.6\" style=\"padding:2.5pt 8.7pt;\">55.5 60.2</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.7.4.7\" style=\"padding:2.5pt 8.7pt;\">\n<span class=\"ltx_text ltx_font_bold\" id=\"S4.T1.1.7.4.7.1\">71.1</span> 60.6</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.7.4.8\" style=\"padding:2.5pt 8.7pt;\">56.5 56.1</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T1.1.8.5\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_r\" id=\"S4.T1.1.8.5.1\" style=\"padding:2.5pt 8.7pt;\">LFM</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.8.5.2\" style=\"padding:2.5pt 8.7pt;\">25.6 33.1</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.8.5.3\" style=\"padding:2.5pt 8.7pt;\">75.1 78.8</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.8.5.4\" style=\"padding:2.5pt 8.7pt;\">58.5 59.2</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.8.5.5\" style=\"padding:2.5pt 8.7pt;\">64.7 65.2</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.8.5.6\" style=\"padding:2.5pt 8.7pt;\">80.2 71.8</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.8.5.7\" style=\"padding:2.5pt 8.7pt;\">61.1 58.9</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.8.5.8\" style=\"padding:2.5pt 8.7pt;\">63.4 62.7</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T1.1.9.6\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_r\" id=\"S4.T1.1.9.6.1\" style=\"padding:2.5pt 8.7pt;\">ICON</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.9.6.2\" style=\"padding:2.5pt 8.7pt;\">22.2 29.9</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.9.6.3\" style=\"padding:2.5pt 8.7pt;\">58.8 64.6</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.9.6.4\" style=\"padding:2.5pt 8.7pt;\">62.8 57.4</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.9.6.5\" style=\"padding:2.5pt 8.7pt;\">64.7 63.0</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.9.6.6\" style=\"padding:2.5pt 8.7pt;\">58.9 63.4</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.9.6.7\" style=\"padding:2.5pt 8.7pt;\">67.2 60.8</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.9.6.8\" style=\"padding:2.5pt 8.7pt;\">59.1 58.5</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T1.1.10.7\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_r\" id=\"S4.T1.1.10.7.1\" style=\"padding:2.5pt 8.7pt;\">A-DMN</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.10.7.2\" style=\"padding:2.5pt 8.7pt;\">43.1 50.6</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.10.7.3\" style=\"padding:2.5pt 8.7pt;\">69.4 76.8</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.10.7.4\" style=\"padding:2.5pt 8.7pt;\">63.0 62.9</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.10.7.5\" style=\"padding:2.5pt 8.7pt;\">63.5 56.5</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.10.7.6\" style=\"padding:2.5pt 8.7pt;\">\n<span class=\"ltx_text ltx_font_bold\" id=\"S4.T1.1.10.7.6.1\">88.3</span> 77.9</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.10.7.7\" style=\"padding:2.5pt 8.7pt;\">53.3 55.7</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.10.7.8\" style=\"padding:2.5pt 8.7pt;\">64.6 64.3</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T1.1.11.8\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_r\" id=\"S4.T1.1.11.8.1\" style=\"padding:2.5pt 8.7pt;\">DialogueGCN</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.11.8.2\" style=\"padding:2.5pt 8.7pt;\">40.6 42.7</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.11.8.3\" style=\"padding:2.5pt 8.7pt;\"><span class=\"ltx_text ltx_font_bold\" id=\"S4.T1.1.11.8.3.1\">89.1 84.5</span></td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.11.8.4\" style=\"padding:2.5pt 8.7pt;\">62.0 63.5</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.11.8.5\" style=\"padding:2.5pt 8.7pt;\">67.5 64.1</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.11.8.6\" style=\"padding:2.5pt 8.7pt;\">65.5 63.1</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.11.8.7\" style=\"padding:2.5pt 8.7pt;\">64.1 66.9</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.11.8.8\" style=\"padding:2.5pt 8.7pt;\">65.2 64.1</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T1.1.12.9\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_r\" id=\"S4.T1.1.12.9.1\" style=\"padding:2.5pt 8.7pt;\">RGAT</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.12.9.2\" style=\"padding:2.5pt 8.7pt;\">60.1 51.6</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.12.9.3\" style=\"padding:2.5pt 8.7pt;\">78.8 77.3</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.12.9.4\" style=\"padding:2.5pt 8.7pt;\">60.1 65.4</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.12.9.5\" style=\"padding:2.5pt 8.7pt;\">70.7 63.0</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.12.9.6\" style=\"padding:2.5pt 8.7pt;\">78.0 68.0</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.12.9.7\" style=\"padding:2.5pt 8.7pt;\">64.3 61.2</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.12.9.8\" style=\"padding:2.5pt 8.7pt;\">65.0 65.2</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T1.1.13.10\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_r\" id=\"S4.T1.1.13.10.1\" style=\"padding:2.5pt 8.7pt;\">AGHMN</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.13.10.2\" style=\"padding:2.5pt 8.7pt;\">48.3 52.1</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.13.10.3\" style=\"padding:2.5pt 8.7pt;\">68.3 73.3</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.13.10.4\" style=\"padding:2.5pt 8.7pt;\">61.6 58.4</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.13.10.5\" style=\"padding:2.5pt 8.7pt;\">57.5 61.9</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.13.10.6\" style=\"padding:2.5pt 8.7pt;\">68.1 69.7</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.13.10.7\" style=\"padding:2.5pt 8.7pt;\">67.1 62.3</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.13.10.8\" style=\"padding:2.5pt 8.7pt;\">63.5 63.5</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T1.1.14.11\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_r\" id=\"S4.T1.1.14.11.1\" style=\"padding:2.5pt 8.7pt;\">BiERU</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.14.11.2\" style=\"padding:2.5pt 8.7pt;\">54.2 31.5</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.14.11.3\" style=\"padding:2.5pt 8.7pt;\">80.6 84.2</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.14.11.4\" style=\"padding:2.5pt 8.7pt;\">64.7 60.2</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.14.11.5\" style=\"padding:2.5pt 8.7pt;\">67.9 65.7</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.14.11.6\" style=\"padding:2.5pt 8.7pt;\">62.8 74.1</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.14.11.7\" style=\"padding:2.5pt 8.7pt;\">61.9 61.3</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.14.11.8\" style=\"padding:2.5pt 8.7pt;\">66.1 64.7</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T1.1.15.12\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_r\" id=\"S4.T1.1.15.12.1\" style=\"padding:2.5pt 8.7pt;\">CoMPM</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.15.12.2\" style=\"padding:2.5pt 8.7pt;\">59.9 60.7</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.15.12.3\" style=\"padding:2.5pt 8.7pt;\">78.0 82.2</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.15.12.4\" style=\"padding:2.5pt 8.7pt;\">60.4 63.0</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.15.12.5\" style=\"padding:2.5pt 8.7pt;\">70.2 59.9</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.15.12.6\" style=\"padding:2.5pt 8.7pt;\">85.8 78.2</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.15.12.7\" style=\"padding:2.5pt 8.7pt;\">62.9 59.5</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.15.12.8\" style=\"padding:2.5pt 8.7pt;\">67.7 67.2</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T1.1.16.13\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_r\" id=\"S4.T1.1.16.13.1\" style=\"padding:2.5pt 8.7pt;\">EmoBERTa</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.16.13.2\" style=\"padding:2.5pt 8.7pt;\">56.9 56.4</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.16.13.3\" style=\"padding:2.5pt 8.7pt;\">79.1 83.0</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.16.13.4\" style=\"padding:2.5pt 8.7pt;\">64.0 61.5</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.16.13.5\" style=\"padding:2.5pt 8.7pt;\">70.6 69.6</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.16.13.6\" style=\"padding:2.5pt 8.7pt;\">86.0 78.0</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.16.13.7\" style=\"padding:2.5pt 8.7pt;\">63.8 68.7</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.16.13.8\" style=\"padding:2.5pt 8.7pt;\">67.3 67.3</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T1.1.17.14\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_r\" id=\"S4.T1.1.17.14.1\" style=\"padding:2.5pt 8.7pt;\">COGMEN</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.17.14.2\" style=\"padding:2.5pt 8.7pt;\">57.4 51.9</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.17.14.3\" style=\"padding:2.5pt 8.7pt;\">81.4 81.7</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.17.14.4\" style=\"padding:2.5pt 8.7pt;\">65.4 <span class=\"ltx_text ltx_font_bold\" id=\"S4.T1.1.17.14.4.1\">68.6</span>\n</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.17.14.5\" style=\"padding:2.5pt 8.7pt;\">69.5 66.0</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.17.14.6\" style=\"padding:2.5pt 8.7pt;\">83.3 75.3</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.17.14.7\" style=\"padding:2.5pt 8.7pt;\">63.8 68.2</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.17.14.8\" style=\"padding:2.5pt 8.7pt;\">68.2 67.6</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T1.1.18.15\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_r\" id=\"S4.T1.1.18.15.1\" style=\"padding:2.5pt 8.7pt;\">CTNet</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.18.15.2\" style=\"padding:2.5pt 8.7pt;\">47.9 51.3</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.18.15.3\" style=\"padding:2.5pt 8.7pt;\">78.0 79.9</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.18.15.4\" style=\"padding:2.5pt 8.7pt;\">\n<span class=\"ltx_text ltx_font_bold\" id=\"S4.T1.1.18.15.4.1\">69.0</span> 65.8</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.18.15.5\" style=\"padding:2.5pt 8.7pt;\">\n<span class=\"ltx_text ltx_font_bold\" id=\"S4.T1.1.18.15.5.1\">72.9</span> 67.2</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.18.15.6\" style=\"padding:2.5pt 8.7pt;\">85.3 78.7</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.18.15.7\" style=\"padding:2.5pt 8.7pt;\">52.2 58.8</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.18.15.8\" style=\"padding:2.5pt 8.7pt;\">68.0 67.5</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T1.1.19.16\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_r\" id=\"S4.T1.1.19.16.1\" style=\"padding:2.5pt 8.7pt;\">LR-GCN</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.19.16.2\" style=\"padding:2.5pt 8.7pt;\">54.2 55.5</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.19.16.3\" style=\"padding:2.5pt 8.7pt;\">81.6 79.1</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.19.16.4\" style=\"padding:2.5pt 8.7pt;\">59.1 63.8</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.19.16.5\" style=\"padding:2.5pt 8.7pt;\">69.4 69.0</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.19.16.6\" style=\"padding:2.5pt 8.7pt;\">76.3 74.0</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.19.16.7\" style=\"padding:2.5pt 8.7pt;\">68.2 <span class=\"ltx_text ltx_font_bold\" id=\"S4.T1.1.19.16.7.1\">68.9</span>\n</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.19.16.8\" style=\"padding:2.5pt 8.7pt;\">68.5 68.3</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T1.1.20.17\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_r\" id=\"S4.T1.1.20.17.1\" style=\"padding:2.5pt 8.7pt;\">DER-GCN</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.20.17.2\" style=\"padding:2.5pt 8.7pt;\">60.7 58.8</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.20.17.3\" style=\"padding:2.5pt 8.7pt;\">75.9 79.8</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.20.17.4\" style=\"padding:2.5pt 8.7pt;\">66.5 61.5</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.20.17.5\" style=\"padding:2.5pt 8.7pt;\">71.3 <span class=\"ltx_text ltx_font_bold\" id=\"S4.T1.1.20.17.5.1\">72.1</span>\n</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.20.17.6\" style=\"padding:2.5pt 8.7pt;\">71.1 73.3</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.20.17.7\" style=\"padding:2.5pt 8.7pt;\">66.1 67.8</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T1.1.20.17.8\" style=\"padding:2.5pt 8.7pt;\">69.7 69.4</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T1.1.21.18\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_b ltx_border_r\" id=\"S4.T1.1.21.18.1\" style=\"padding:2.5pt 8.7pt;\">ELR-GCN</th>\n<td class=\"ltx_td ltx_align_center ltx_border_b\" id=\"S4.T1.1.21.18.2\" style=\"padding:2.5pt 8.7pt;\"><span class=\"ltx_text ltx_font_bold\" id=\"S4.T1.1.21.18.2.1\">64.7 62.9</span></td>\n<td class=\"ltx_td ltx_align_center ltx_border_b\" id=\"S4.T1.1.21.18.3\" style=\"padding:2.5pt 8.7pt;\">75.7 80.8</td>\n<td class=\"ltx_td ltx_align_center ltx_border_b\" id=\"S4.T1.1.21.18.4\" style=\"padding:2.5pt 8.7pt;\">66.2 62.4</td>\n<td class=\"ltx_td ltx_align_center ltx_border_b\" id=\"S4.T1.1.21.18.5\" style=\"padding:2.5pt 8.7pt;\">70.7 70.0</td>\n<td class=\"ltx_td ltx_align_center ltx_border_b\" id=\"S4.T1.1.21.18.6\" style=\"padding:2.5pt 8.7pt;\">76.8 <span class=\"ltx_text ltx_font_bold\" id=\"S4.T1.1.21.18.6.1\">78.6</span>\n</td>\n<td class=\"ltx_td ltx_align_center ltx_border_b\" id=\"S4.T1.1.21.18.7\" style=\"padding:2.5pt 8.7pt;\">67.9 68.1</td>\n<td class=\"ltx_td ltx_align_center ltx_border_b\" id=\"S4.T1.1.21.18.8\" style=\"padding:2.5pt 8.7pt;\"><span class=\"ltx_text ltx_font_bold\" id=\"S4.T1.1.21.18.8.1\">70.6 70.9</span></td>\n</tr>\n</tbody>\n</table>\n</figure>",
        "perturb_sentence_id": [
            2,
            3
        ],
        "output": {
            "perturbed_statement": "[paragraph id = 2] IEMOCAP: As shown in Table 1, the multi-modal emotion recognition method proposed in this paper achieved the best emotion recognition effect on the IEMOCAP data set, with an average accuracy of 70.6% and an average F1 value of 70.9%. However, DER-GCN proposes an effective modeling method of long-distance context latent dependencies for multi-modal emotion recognition.",
            "perturbed_explanation": "1. ELR-GCN proposes an effective modeling method of long-distance context latent dependencies for multi-modal emotion recognition. 2. The statement incorrectly attributes the method of modeling long-distance context latent dependencies to DER-GCN, whereas it is actually ELR-GCN that proposes this effective modeling method."
        }
    },
    {
        "path": "table_paper/2407.00119v2.json",
        "table_id": "2",
        "section": "4.3",
        "all_context": [
            "To verify the superiority of the ELR-GNN method proposed in this paper, we report the experimental results of ELR-GNN and other comparative methods on the IEMOCAP and MELD data sets.",
            "Experimental results are presented in Tables 1 and 2 .",
            "IEMOCAP: As shown in Table 1 , the multi-modal emotion recognition method proposed in this paper achie-ved the best emotion recognition effect on the IEMOCAP data set, with an average accuracy of 70.6% and an average F1 value of 70.9%.",
            "ELR-GCN proposes an effective modeling method of long-distance context latent dependencies for multi-modal emotion recognition.",
            "In addition, ELR-GCN also combines early and adaptive late fusion methods to achieve the capture of fine-grained emotional features.",
            "Among other comparison methods, the emotion recognition effect of DER-GCN is slightly lower than that of ELR-GNN, with an average accuracy of 69.7% and an average F1 value of 69.4%.",
            "Although DER-GCN comprehensively considers event relationships and dialogue relationships between speakers to enhance the model s emotional understanding, it ignores latent context dependencies.",
            "The emotion recognition effect of LR-GCN is lower than ELR-GNN and DER-GCN, with an average accuracy of 68.5% and an average F1 value of 68.3%.",
            "Although LR-GCN considers latent dependencies between contexts, due to the high computational complexity of GCN, LR-GCN can only capture local latent dependencies.",
            "The emotion recognition effects of other comparison methods are lower than ELR-GNN.",
            "Likewise, none of them take into account potential dependencies on context.",
            "Overall, the accuracy of ELR-GNN on the happy emotion analogy is much higher than that of other comparison algorithms, while the accuracy of other emotion categories is also relatively close to that of other comparison algorithms.",
            "In addition, the F1 value of ELR-GNN on the happy and excited emotional analogies is much higher than that of other comparison algorithms.",
            "At the same time, the F1 value of ELR-GNN on other emotional categories is also relatively close to other comparison algorithms.",
            "The experimental results prove the superiority of the ELR-GNN method proposed in this paper.",
            "MELD: As shown in Table 2 , The ELR-GNN method proposed in this article has the best emotion recognition effect on the MELD data set, with an average accuracy of 68.7% and an average F1 value of 69.9%.",
            "The emotion recognition effect of DER-GCN is second, with an average accuracy of 69.7% and an average F1 value of 69.4%.",
            "The emotion recognition effect of LR-GCN is lower than that of ELR-GNN and DER-GCN, with an average accuracy of 68.5% and an average F1 value of 68.3%.",
            "The emotion recognition effects of other comparison methods are relatively poor, and the average accuracy and F1 value are lower than ELR-GNN.",
            "The performance improvement may be attributed to ELR-GNN s ability to capture long-distance contextual latent dependencies and fine-grained fusion of dialogue relationships between speakers, contextual latent dependencies and contextual semantic information.",
            "Overall, the accuracy of ELR-GNN on the neutral, fear, sadness, joy, and disgust emotion analogy is much higher than that of other comparison algorithms, while the accuracy of other emotion categories is also relatively close to that of other comparison algorithms.",
            "In addition, the F1 value of ELR-GNN on the neutral, fear, sadness, joy, and anger emotional analogies is much higher than that of other comparison algorithms.",
            "At the same time, the F1 value of ELR-GNN on other emotional categories is also relatively close to other comparison algorithms.",
            "In addition, we find that ELR-GNN has better emotion recognition effects on the minority emotions fear and disgust, with relatively high accuracy and F1 value.",
            "The experimental results prove the superiority of the ELR-GNN method proposed in this paper.",
            "In addition, to intuitively illustrate that the running time of the ELR-GNN method proposed in this paper is better than other comparative methods, we statistics in Table 3 the running time of other comparative methods of the ELR-GNN method on the IEMOCAP and MELD data sets.",
            "As shown in Table 3 , the running time of the ELR-GNN method proposed in this paper on the IEMOCAP and MELD data sets is 41s and 91s respectively, which is significantly better than other comparison methods.",
            "The running times of DialogueGCN are 58s and 127s respectively, which are lower than LR-GCN and DER-GCN, but the emotion recognition effect is relatively poor.",
            "The running times of LR-GCN are 87s and 142s respectively.",
            "The running times of DER-GCN are 125s and 189s respectively.",
            "The experimental results prove the efficiency and effectiveness of the ELR-GNN method proposed in this paper.",
            ""
        ],
        "target_context_ids": [
            1,
            15,
            16,
            17,
            18,
            19,
            20,
            21,
            22,
            23,
            24,
            25
        ],
        "selected_paragraphs": [
            "[paragraph id = 1] Experimental results are presented in Tables 1 and 2 .",
            "[paragraph id = 15] MELD: As shown in Table 2 , The ELR-GNN method proposed in this article has the best emotion recognition effect on the MELD data set, with an average accuracy of 68.7% and an average F1 value of 69.9%.",
            "[paragraph id = 16] The emotion recognition effect of DER-GCN is second, with an average accuracy of 69.7% and an average F1 value of 69.4%.",
            "[paragraph id = 17] The emotion recognition effect of LR-GCN is lower than that of ELR-GNN and DER-GCN, with an average accuracy of 68.5% and an average F1 value of 68.3%.",
            "[paragraph id = 18] The emotion recognition effects of other comparison methods are relatively poor, and the average accuracy and F1 value are lower than ELR-GNN.",
            "[paragraph id = 19] The performance improvement may be attributed to ELR-GNN s ability to capture long-distance contextual latent dependencies and fine-grained fusion of dialogue relationships between speakers, contextual latent dependencies and contextual semantic information.",
            "[paragraph id = 20] Overall, the accuracy of ELR-GNN on the neutral, fear, sadness, joy, and disgust emotion analogy is much higher than that of other comparison algorithms, while the accuracy of other emotion categories is also relatively close to that of other comparison algorithms.",
            "[paragraph id = 21] In addition, the F1 value of ELR-GNN on the neutral, fear, sadness, joy, and anger emotional analogies is much higher than that of other comparison algorithms.",
            "[paragraph id = 22] At the same time, the F1 value of ELR-GNN on other emotional categories is also relatively close to other comparison algorithms.",
            "[paragraph id = 23] In addition, we find that ELR-GNN has better emotion recognition effects on the minority emotions fear and disgust, with relatively high accuracy and F1 value.",
            "[paragraph id = 24] The experimental results prove the superiority of the ELR-GNN method proposed in this paper.",
            "[paragraph id = 25] In addition, to intuitively illustrate that the running time of the ELR-GNN method proposed in this paper is better than other comparative methods, we statistics in Table 3 the running time of other comparative methods of the ELR-GNN method on the IEMOCAP and MELD data sets."
        ],
        "table_html": "<figure class=\"ltx_table\" id=\"S4.T2\">\n<figcaption class=\"ltx_caption\"><span class=\"ltx_tag ltx_tag_table\">Table 2: </span>Comparison with other baseline models on the MELD dataset.</figcaption>\n<table class=\"ltx_tabular ltx_guessed_headers ltx_align_middle\" id=\"S4.T2.1\">\n<thead class=\"ltx_thead\">\n<tr class=\"ltx_tr\" id=\"S4.T2.1.1.1\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_column ltx_th_row ltx_border_r ltx_border_t\" id=\"S4.T2.1.1.1.1\" rowspan=\"3\" style=\"padding:2.5pt 5.4pt;\"><span class=\"ltx_text\" id=\"S4.T2.1.1.1.1.1\">Methods</span></th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" colspan=\"8\" id=\"S4.T2.1.1.1.2\" style=\"padding:2.5pt 5.4pt;\">MELD</th>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T2.1.2.2\">\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T2.1.2.2.1\" style=\"padding:2.5pt 5.4pt;\">Neutral</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T2.1.2.2.2\" style=\"padding:2.5pt 5.4pt;\">Surprise</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T2.1.2.2.3\" style=\"padding:2.5pt 5.4pt;\">Fear</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T2.1.2.2.4\" style=\"padding:2.5pt 5.4pt;\">Sadness</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T2.1.2.2.5\" style=\"padding:2.5pt 5.4pt;\">Joy</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T2.1.2.2.6\" style=\"padding:2.5pt 5.4pt;\">Disgust</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T2.1.2.2.7\" style=\"padding:2.5pt 5.4pt;\">Anger</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T2.1.2.2.8\" style=\"padding:2.5pt 5.4pt;\">Average(w)</th>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T2.1.3.3\">\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T2.1.3.3.1\" style=\"padding:2.5pt 5.4pt;\">Acc. F1</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T2.1.3.3.2\" style=\"padding:2.5pt 5.4pt;\">Acc. F1</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T2.1.3.3.3\" style=\"padding:2.5pt 5.4pt;\">Acc. F1</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T2.1.3.3.4\" style=\"padding:2.5pt 5.4pt;\">Acc. F1</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T2.1.3.3.5\" style=\"padding:2.5pt 5.4pt;\">Acc. F1</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T2.1.3.3.6\" style=\"padding:2.5pt 5.4pt;\">Acc. F1</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T2.1.3.3.7\" style=\"padding:2.5pt 5.4pt;\">Acc. F1</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T2.1.3.3.8\" style=\"padding:2.5pt 5.4pt;\">Acc. F1</th>\n</tr>\n</thead>\n<tbody class=\"ltx_tbody\">\n<tr class=\"ltx_tr\" id=\"S4.T2.1.4.1\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_r ltx_border_t\" id=\"S4.T2.1.4.1.1\" style=\"padding:2.5pt 5.4pt;\">TextCNN</th>\n<td class=\"ltx_td ltx_align_center ltx_border_t\" id=\"S4.T2.1.4.1.2\" style=\"padding:2.5pt 5.4pt;\">76.2 74.9</td>\n<td class=\"ltx_td ltx_align_center ltx_border_t\" id=\"S4.T2.1.4.1.3\" style=\"padding:2.5pt 5.4pt;\">43.3 45.5</td>\n<td class=\"ltx_td ltx_align_center ltx_border_t\" id=\"S4.T2.1.4.1.4\" style=\"padding:2.5pt 5.4pt;\">4.6 3.7</td>\n<td class=\"ltx_td ltx_align_center ltx_border_t\" id=\"S4.T2.1.4.1.5\" style=\"padding:2.5pt 5.4pt;\">18.2 21.1</td>\n<td class=\"ltx_td ltx_align_center ltx_border_t\" id=\"S4.T2.1.4.1.6\" style=\"padding:2.5pt 5.4pt;\">46.1 49.4</td>\n<td class=\"ltx_td ltx_align_center ltx_border_t\" id=\"S4.T2.1.4.1.7\" style=\"padding:2.5pt 5.4pt;\">8.9 8.3</td>\n<td class=\"ltx_td ltx_align_center ltx_border_t\" id=\"S4.T2.1.4.1.8\" style=\"padding:2.5pt 5.4pt;\">35.3 34.5</td>\n<td class=\"ltx_td ltx_align_center ltx_border_t\" id=\"S4.T2.1.4.1.9\" style=\"padding:2.5pt 5.4pt;\">56.3 55.0</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T2.1.5.2\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_r\" id=\"S4.T2.1.5.2.1\" style=\"padding:2.5pt 5.4pt;\">bc-LSTM</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.5.2.2\" style=\"padding:2.5pt 5.4pt;\">78.4 73.8</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.5.2.3\" style=\"padding:2.5pt 5.4pt;\">46.8 47.7</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.5.2.4\" style=\"padding:2.5pt 5.4pt;\">3.8 5.4</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.5.2.5\" style=\"padding:2.5pt 5.4pt;\">22.4 25.1</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.5.2.6\" style=\"padding:2.5pt 5.4pt;\">51.6 51.3</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.5.2.7\" style=\"padding:2.5pt 5.4pt;\">4.3 5.2</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.5.2.8\" style=\"padding:2.5pt 5.4pt;\">36.7 38.4</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.5.2.9\" style=\"padding:2.5pt 5.4pt;\">57.5 55.9</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T2.1.6.3\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_r\" id=\"S4.T2.1.6.3.1\" style=\"padding:2.5pt 5.4pt;\">DialogueRNN</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.6.3.2\" style=\"padding:2.5pt 5.4pt;\">72.1 73.5</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.6.3.3\" style=\"padding:2.5pt 5.4pt;\">54.4 49.4</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.6.3.4\" style=\"padding:2.5pt 5.4pt;\">1.6 1.2</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.6.3.5\" style=\"padding:2.5pt 5.4pt;\">23.9 23.8</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.6.3.6\" style=\"padding:2.5pt 5.4pt;\">52.0 50.7</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.6.3.7\" style=\"padding:2.5pt 5.4pt;\">1.5 1.7</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.6.3.8\" style=\"padding:2.5pt 5.4pt;\">41.0 41.5</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.6.3.9\" style=\"padding:2.5pt 5.4pt;\">56.1 55.9</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T2.1.7.4\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_r\" id=\"S4.T2.1.7.4.1\" style=\"padding:2.5pt 5.4pt;\">DialogueGCN</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.7.4.2\" style=\"padding:2.5pt 5.4pt;\">70.3 72.1</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.7.4.3\" style=\"padding:2.5pt 5.4pt;\">42.4 41.7</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.7.4.4\" style=\"padding:2.5pt 5.4pt;\">3.0 2.8</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.7.4.5\" style=\"padding:2.5pt 5.4pt;\">20.9 21.8</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.7.4.6\" style=\"padding:2.5pt 5.4pt;\">44.7 44.2</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.7.4.7\" style=\"padding:2.5pt 5.4pt;\">6.5 6.7</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.7.4.8\" style=\"padding:2.5pt 5.4pt;\">39.0 36.5</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.7.4.9\" style=\"padding:2.5pt 5.4pt;\">54.9 54.7</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T2.1.8.5\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_r\" id=\"S4.T2.1.8.5.1\" style=\"padding:2.5pt 5.4pt;\">RGAT</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.8.5.2\" style=\"padding:2.5pt 5.4pt;\">76.0 78.1</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.8.5.3\" style=\"padding:2.5pt 5.4pt;\">40.1 41.5</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.8.5.4\" style=\"padding:2.5pt 5.4pt;\">3.0 2.4</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.8.5.5\" style=\"padding:2.5pt 5.4pt;\">32.1 30.7</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.8.5.6\" style=\"padding:2.5pt 5.4pt;\">68.1 58.6</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.8.5.7\" style=\"padding:2.5pt 5.4pt;\">4.5 2.2</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.8.5.8\" style=\"padding:2.5pt 5.4pt;\">40.0 44.6</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.8.5.9\" style=\"padding:2.5pt 5.4pt;\">60.3 61.1</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T2.1.9.6\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_r\" id=\"S4.T2.1.9.6.1\" style=\"padding:2.5pt 5.4pt;\">CoMPM</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.9.6.2\" style=\"padding:2.5pt 5.4pt;\">78.3 82.0</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.9.6.3\" style=\"padding:2.5pt 5.4pt;\">48.3 49.2</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.9.6.4\" style=\"padding:2.5pt 5.4pt;\">1.7 2.9</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.9.6.5\" style=\"padding:2.5pt 5.4pt;\">35.9 32.3</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.9.6.6\" style=\"padding:2.5pt 5.4pt;\">71.4 61.5</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.9.6.7\" style=\"padding:2.5pt 5.4pt;\">3.1 2.8</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.9.6.8\" style=\"padding:2.5pt 5.4pt;\">42.2 45.8</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.9.6.9\" style=\"padding:2.5pt 5.4pt;\">64.1 65.3</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T2.1.10.7\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_r\" id=\"S4.T2.1.10.7.1\" style=\"padding:2.5pt 5.4pt;\">EmoBERTa</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.10.7.2\" style=\"padding:2.5pt 5.4pt;\"><span class=\"ltx_text ltx_font_bold\" id=\"S4.T2.1.10.7.2.1\">78.9 82.5</span></td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.10.7.3\" style=\"padding:2.5pt 5.4pt;\">50.2 50.2</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.10.7.4\" style=\"padding:2.5pt 5.4pt;\">1.8 1.9</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.10.7.5\" style=\"padding:2.5pt 5.4pt;\">33.3 31.2</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.10.7.6\" style=\"padding:2.5pt 5.4pt;\">72.1 61.7</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.10.7.7\" style=\"padding:2.5pt 5.4pt;\">9.1 2.5</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.10.7.8\" style=\"padding:2.5pt 5.4pt;\">43.3 46.4</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.10.7.9\" style=\"padding:2.5pt 5.4pt;\">64.1 65.2</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T2.1.11.8\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_r\" id=\"S4.T2.1.11.8.1\" style=\"padding:2.5pt 5.4pt;\">ConGCN</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.11.8.2\" style=\"padding:2.5pt 5.4pt;\">46.8 45.4</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.11.8.3\" style=\"padding:2.5pt 5.4pt;\">10.6 8.8</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.11.8.4\" style=\"padding:2.5pt 5.4pt;\">8.7 8.1</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.11.8.5\" style=\"padding:2.5pt 5.4pt;\">53.1 54.6</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.11.8.6\" style=\"padding:2.5pt 5.4pt;\">76.7 75.2</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.11.8.7\" style=\"padding:2.5pt 5.4pt;\">28.5 <span class=\"ltx_text ltx_font_bold\" id=\"S4.T2.1.11.8.7.1\">26.3</span>\n</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.11.8.8\" style=\"padding:2.5pt 5.4pt;\">50.3 48.4</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.11.8.9\" style=\"padding:2.5pt 5.4pt;\">59.4 58.7</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T2.1.12.9\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_r\" id=\"S4.T2.1.12.9.1\" style=\"padding:2.5pt 5.4pt;\">A-DMN</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.12.9.2\" style=\"padding:2.5pt 5.4pt;\">76.5 78.9</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.12.9.3\" style=\"padding:2.5pt 5.4pt;\"><span class=\"ltx_text ltx_font_bold\" id=\"S4.T2.1.12.9.3.1\">56.2 55.3</span></td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.12.9.4\" style=\"padding:2.5pt 5.4pt;\">8.2 8.6</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.12.9.5\" style=\"padding:2.5pt 5.4pt;\">22.1 24.9</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.12.9.6\" style=\"padding:2.5pt 5.4pt;\">59.8 57.4</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.12.9.7\" style=\"padding:2.5pt 5.4pt;\">1.2 3.4</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.12.9.8\" style=\"padding:2.5pt 5.4pt;\">41.3 40.9</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.12.9.9\" style=\"padding:2.5pt 5.4pt;\">61.5 60.4</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T2.1.13.10\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_r\" id=\"S4.T2.1.13.10.1\" style=\"padding:2.5pt 5.4pt;\">LR-GCN</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.13.10.2\" style=\"padding:2.5pt 5.4pt;\">76.7 80.0</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.13.10.3\" style=\"padding:2.5pt 5.4pt;\">53.3 55.2</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.13.10.4\" style=\"padding:2.5pt 5.4pt;\">0.0 0.0</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.13.10.5\" style=\"padding:2.5pt 5.4pt;\">49.6 35.1</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.13.10.6\" style=\"padding:2.5pt 5.4pt;\">68.0 64.4</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.13.10.7\" style=\"padding:2.5pt 5.4pt;\">10.7 2.7</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.13.10.8\" style=\"padding:2.5pt 5.4pt;\">48.0 51.0</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.13.10.9\" style=\"padding:2.5pt 5.4pt;\">65.7 65.6</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T2.1.14.11\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_r\" id=\"S4.T2.1.14.11.1\" style=\"padding:2.5pt 5.4pt;\">DER-GCN</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.14.11.2\" style=\"padding:2.5pt 5.4pt;\">76.8 80.6</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.14.11.3\" style=\"padding:2.5pt 5.4pt;\">50.5 51.0</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.14.11.4\" style=\"padding:2.5pt 5.4pt;\">14.8 10.4</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.14.11.5\" style=\"padding:2.5pt 5.4pt;\">56.7 41.5</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.14.11.6\" style=\"padding:2.5pt 5.4pt;\">69.3 64.3</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.14.11.7\" style=\"padding:2.5pt 5.4pt;\">17.2 10.3</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.14.11.8\" style=\"padding:2.5pt 5.4pt;\">\n<span class=\"ltx_text ltx_font_bold\" id=\"S4.T2.1.14.11.8.1\">52.5</span> 57.4</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T2.1.14.11.9\" style=\"padding:2.5pt 5.4pt;\">66.8 66.1</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T2.1.15.12\">\n<th class=\"ltx_td ltx_align_left ltx_th ltx_th_row ltx_border_b ltx_border_r\" id=\"S4.T2.1.15.12.1\" style=\"padding:2.5pt 5.4pt;\">ELR-GCN</th>\n<td class=\"ltx_td ltx_align_center ltx_border_b\" id=\"S4.T2.1.15.12.2\" style=\"padding:2.5pt 5.4pt;\"><span class=\"ltx_text ltx_font_bold\" id=\"S4.T2.1.15.12.2.1\">80.2 83.6</span></td>\n<td class=\"ltx_td ltx_align_center ltx_border_b\" id=\"S4.T2.1.15.12.3\" style=\"padding:2.5pt 5.4pt;\">36.8 35.4</td>\n<td class=\"ltx_td ltx_align_center ltx_border_b\" id=\"S4.T2.1.15.12.4\" style=\"padding:2.5pt 5.4pt;\"><span class=\"ltx_text ltx_font_bold\" id=\"S4.T2.1.15.12.4.1\">19.2 13.1</span></td>\n<td class=\"ltx_td ltx_align_center ltx_border_b\" id=\"S4.T2.1.15.12.5\" style=\"padding:2.5pt 5.4pt;\"><span class=\"ltx_text ltx_font_bold\" id=\"S4.T2.1.15.12.5.1\">80.2 83.6</span></td>\n<td class=\"ltx_td ltx_align_center ltx_border_b\" id=\"S4.T2.1.15.12.6\" style=\"padding:2.5pt 5.4pt;\"><span class=\"ltx_text ltx_font_bold\" id=\"S4.T2.1.15.12.6.1\">76.5 69.7</span></td>\n<td class=\"ltx_td ltx_align_center ltx_border_b\" id=\"S4.T2.1.15.12.7\" style=\"padding:2.5pt 5.4pt;\">\n<span class=\"ltx_text ltx_font_bold\" id=\"S4.T2.1.15.12.7.1\">55.6</span> 13.0</td>\n<td class=\"ltx_td ltx_align_center ltx_border_b\" id=\"S4.T2.1.15.12.8\" style=\"padding:2.5pt 5.4pt;\">52.1 <span class=\"ltx_text ltx_font_bold\" id=\"S4.T2.1.15.12.8.1\">57.7</span>\n</td>\n<td class=\"ltx_td ltx_align_center ltx_border_b\" id=\"S4.T2.1.15.12.9\" style=\"padding:2.5pt 5.4pt;\"><span class=\"ltx_text ltx_font_bold\" id=\"S4.T2.1.15.12.9.1\">68.7 69.9</span></td>\n</tr>\n</tbody>\n</table>\n</figure>",
        "perturb_sentence_id": [
            15,
            16
        ],
        "output": {
            "perturbed_statement": "[paragraph id = 15] MELD: As shown in Table 2, The ELR-GNN method proposed in this article has the best emotion recognition effect on the MELD data set, with an average accuracy of 68.7% and an average F1 value of 69.9%. The emotion recognition effect of DER-GCN is second, with an average accuracy of 68.5% and an average F1 value of 68.3%.",
            "perturbed_explanation": "1. The ELR-GNN method is stated to have the best emotion recognition effect on the MELD data set in terms of its performance metrics. 2. The statement claims that DER-GCN has an average accuracy of 68.5% and an average F1 value of 68.3%, which are actually the metrics mentioned in the context for LR-GCN, not DER-GCN. DER-GCN's performance metrics are not provided in the context."
        }
    },
    {
        "path": "table_paper/2407.00119v2.json",
        "table_id": "3",
        "section": "4.3",
        "all_context": [
            "To verify the superiority of the ELR-GNN method proposed in this paper, we report the experimental results of ELR-GNN and other comparative methods on the IEMOCAP and MELD data sets.",
            "Experimental results are presented in Tables 1 and 2 .",
            "IEMOCAP: As shown in Table 1 , the multi-modal emotion recognition method proposed in this paper achie-ved the best emotion recognition effect on the IEMOCAP data set, with an average accuracy of 70.6% and an average F1 value of 70.9%.",
            "ELR-GCN proposes an effective modeling method of long-distance context latent dependencies for multi-modal emotion recognition.",
            "In addition, ELR-GCN also combines early and adaptive late fusion methods to achieve the capture of fine-grained emotional features.",
            "Among other comparison methods, the emotion recognition effect of DER-GCN is slightly lower than that of ELR-GNN, with an average accuracy of 69.7% and an average F1 value of 69.4%.",
            "Although DER-GCN comprehensively considers event relationships and dialogue relationships between speakers to enhance the model s emotional understanding, it ignores latent context dependencies.",
            "The emotion recognition effect of LR-GCN is lower than ELR-GNN and DER-GCN, with an average accuracy of 68.5% and an average F1 value of 68.3%.",
            "Although LR-GCN considers latent dependencies between contexts, due to the high computational complexity of GCN, LR-GCN can only capture local latent dependencies.",
            "The emotion recognition effects of other comparison methods are lower than ELR-GNN.",
            "Likewise, none of them take into account potential dependencies on context.",
            "Overall, the accuracy of ELR-GNN on the happy emotion analogy is much higher than that of other comparison algorithms, while the accuracy of other emotion categories is also relatively close to that of other comparison algorithms.",
            "In addition, the F1 value of ELR-GNN on the happy and excited emotional analogies is much higher than that of other comparison algorithms.",
            "At the same time, the F1 value of ELR-GNN on other emotional categories is also relatively close to other comparison algorithms.",
            "The experimental results prove the superiority of the ELR-GNN method proposed in this paper.",
            "MELD: As shown in Table 2 , The ELR-GNN method proposed in this article has the best emotion recognition effect on the MELD data set, with an average accuracy of 68.7% and an average F1 value of 69.9%.",
            "The emotion recognition effect of DER-GCN is second, with an average accuracy of 69.7% and an average F1 value of 69.4%.",
            "The emotion recognition effect of LR-GCN is lower than that of ELR-GNN and DER-GCN, with an average accuracy of 68.5% and an average F1 value of 68.3%.",
            "The emotion recognition effects of other comparison methods are relatively poor, and the average accuracy and F1 value are lower than ELR-GNN.",
            "The performance improvement may be attributed to ELR-GNN s ability to capture long-distance contextual latent dependencies and fine-grained fusion of dialogue relationships between speakers, contextual latent dependencies and contextual semantic information.",
            "Overall, the accuracy of ELR-GNN on the neutral, fear, sadness, joy, and disgust emotion analogy is much higher than that of other comparison algorithms, while the accuracy of other emotion categories is also relatively close to that of other comparison algorithms.",
            "In addition, the F1 value of ELR-GNN on the neutral, fear, sadness, joy, and anger emotional analogies is much higher than that of other comparison algorithms.",
            "At the same time, the F1 value of ELR-GNN on other emotional categories is also relatively close to other comparison algorithms.",
            "In addition, we find that ELR-GNN has better emotion recognition effects on the minority emotions fear and disgust, with relatively high accuracy and F1 value.",
            "The experimental results prove the superiority of the ELR-GNN method proposed in this paper.",
            "In addition, to intuitively illustrate that the running time of the ELR-GNN method proposed in this paper is better than other comparative methods, we statistics in Table 3 the running time of other comparative methods of the ELR-GNN method on the IEMOCAP and MELD data sets.",
            "As shown in Table 3 , the running time of the ELR-GNN method proposed in this paper on the IEMOCAP and MELD data sets is 41s and 91s respectively, which is significantly better than other comparison methods.",
            "The running times of DialogueGCN are 58s and 127s respectively, which are lower than LR-GCN and DER-GCN, but the emotion recognition effect is relatively poor.",
            "The running times of LR-GCN are 87s and 142s respectively.",
            "The running times of DER-GCN are 125s and 189s respectively.",
            "The experimental results prove the efficiency and effectiveness of the ELR-GNN method proposed in this paper.",
            ""
        ],
        "target_context_ids": [
            25,
            26,
            27,
            28,
            29,
            30
        ],
        "selected_paragraphs": [
            "[paragraph id = 25] In addition, to intuitively illustrate that the running time of the ELR-GNN method proposed in this paper is better than other comparative methods, we statistics in Table 3 the running time of other comparative methods of the ELR-GNN method on the IEMOCAP and MELD data sets.",
            "[paragraph id = 26] As shown in Table 3 , the running time of the ELR-GNN method proposed in this paper on the IEMOCAP and MELD data sets is 41s and 91s respectively, which is significantly better than other comparison methods.",
            "[paragraph id = 27] The running times of DialogueGCN are 58s and 127s respectively, which are lower than LR-GCN and DER-GCN, but the emotion recognition effect is relatively poor.",
            "[paragraph id = 28] The running times of LR-GCN are 87s and 142s respectively.",
            "[paragraph id = 29] The running times of DER-GCN are 125s and 189s respectively.",
            "[paragraph id = 30] The experimental results prove the efficiency and effectiveness of the ELR-GNN method proposed in this paper."
        ],
        "table_html": "<figure class=\"ltx_table\" id=\"S4.T3\">\n<figcaption class=\"ltx_caption\"><span class=\"ltx_tag ltx_tag_table\">Table 3: </span>We tested the running time of the ELR-GNN method proposed in this paper and other comparative methods on the IEMOCAP and MELD data sets. In particular, ELR-GNN sets  to  and neighbor size to 64.</figcaption>\n<table class=\"ltx_tabular ltx_guessed_headers ltx_align_middle\" id=\"S4.T3.5\">\n<thead class=\"ltx_thead\">\n<tr class=\"ltx_tr\" id=\"S4.T3.5.1.1\">\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_th_row ltx_border_r ltx_border_t\" id=\"S4.T3.5.1.1.1\" rowspan=\"2\" style=\"padding:2.5pt 18.5pt;\">      <span class=\"ltx_text\" id=\"S4.T3.5.1.1.1.1\">Methods</span></th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" colspan=\"2\" id=\"S4.T3.5.1.1.2\" style=\"padding:2.5pt 18.5pt;\">      Running time (s)</th>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T3.5.2.2\">\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T3.5.2.2.1\" style=\"padding:2.5pt 18.5pt;\">      IEMOCAP</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T3.5.2.2.2\" style=\"padding:2.5pt 18.5pt;\">      MELD</th>\n</tr>\n</thead>\n<tbody class=\"ltx_tbody\">\n<tr class=\"ltx_tr\" id=\"S4.T3.5.3.1\">\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_row ltx_border_r ltx_border_t\" id=\"S4.T3.5.3.1.1\" style=\"padding:2.5pt 18.5pt;\">      DialogueGCN</th>\n<td class=\"ltx_td ltx_align_center ltx_border_t\" id=\"S4.T3.5.3.1.2\" style=\"padding:2.5pt 18.5pt;\">      58</td>\n<td class=\"ltx_td ltx_align_center ltx_border_t\" id=\"S4.T3.5.3.1.3\" style=\"padding:2.5pt 18.5pt;\">      127</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T3.5.4.2\">\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_row ltx_border_r\" id=\"S4.T3.5.4.2.1\" style=\"padding:2.5pt 18.5pt;\">      LR-GCN</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T3.5.4.2.2\" style=\"padding:2.5pt 18.5pt;\">      87</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T3.5.4.2.3\" style=\"padding:2.5pt 18.5pt;\">      142</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T3.5.5.3\">\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_row ltx_border_r\" id=\"S4.T3.5.5.3.1\" style=\"padding:2.5pt 18.5pt;\">      DER-GCN</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T3.5.5.3.2\" style=\"padding:2.5pt 18.5pt;\">      125</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T3.5.5.3.3\" style=\"padding:2.5pt 18.5pt;\">      189</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T3.5.6.4\">\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_row ltx_border_b ltx_border_r\" id=\"S4.T3.5.6.4.1\" style=\"padding:2.5pt 18.5pt;\">      ELR-GNN</th>\n<td class=\"ltx_td ltx_align_center ltx_border_b\" id=\"S4.T3.5.6.4.2\" style=\"padding:2.5pt 18.5pt;\">      41</td>\n<td class=\"ltx_td ltx_align_center ltx_border_b\" id=\"S4.T3.5.6.4.3\" style=\"padding:2.5pt 18.5pt;\">      91</td>\n</tr>\n</tbody>\n</table>\n</figure>",
        "perturb_sentence_id": [
            29,
            30
        ],
        "output": {
            "perturbed_statement": "[paragraph id = 29] The running times of DER-GCN are 100s and 150s respectively. The experimental results prove the efficiency and effectiveness of the ELR-GNN method proposed in this paper.",
            "perturbed_explanation": "1. The original explanation asserted that the running times of DER-GCN are 125s and 189s respectively, which aligns with the factual data that could have been provided in the surrounding context about DER-GCN. 2. The statement is incorrect because it claims the running times of DER-GCN to be 100s and 150s respectively, while these figures are not given in the context. Such times do not match with any running times documented for DER-GCN or any other method detailed in the context, which challenges both accuracy and verification of this claim."
        }
    },
    {
        "path": "table_paper/2407.00119v2.json",
        "table_id": "4",
        "section": "4.5.1",
        "all_context": [
            "To verify the importance of the three modal features of text, video and audio for ELR-GNN, we conducted ablation experiments on the IEMOCAP and MELD data sets to compare the performance of the combination of different modal features.",
            "The experimental results are shown in Table 4 .",
            "In single-modal experiments, ELR-GNN with text modality features has the best emotion recognition effect.",
            "The average accuracy on the IEMOCAP and MELD data sets are 64.1% and 63.5%, respectively, and the average F1 value is 63.9% and 62.4%, respectively.",
            "The emotion recognition effect of ELR-GNN with audio modal features is second, with average accuracy rates of 61.1% and 62.7% on the IEMOCAP and MELD data sets, and average F1 values of 60.8% and 62.0% respectively.",
            "ELR-GNN with video modality features has the worst emotion recognition effect, with average accuracy rates of 59.4% and 60.1% on the IEMOCAP and MELD data sets, and average F1 values of 59.7% and 61.4% respectively.",
            "Experimental results show that text features contain the most emotional semantic information.",
            "In the dual-modal experiment, ELR-GNN with text and audio modal features has the best emotion recognition effect.",
            "The average accuracy on the IEMOCAP and MELD data sets are 65.0% and 64.1%, respectively, and the average F1 values are are 64.4% and 63.2%, respectively.",
            "Experimental results demonstrate the effectiveness of multimodal features.",
            ""
        ],
        "target_context_ids": [
            1,
            2,
            3,
            4,
            5,
            6,
            7,
            8,
            9
        ],
        "selected_paragraphs": [
            "[paragraph id = 1] The experimental results are shown in Table 4 .",
            "[paragraph id = 2] In single-modal experiments, ELR-GNN with text modality features has the best emotion recognition effect.",
            "[paragraph id = 3] The average accuracy on the IEMOCAP and MELD data sets are 64.1% and 63.5%, respectively, and the average F1 value is 63.9% and 62.4%, respectively.",
            "[paragraph id = 4] The emotion recognition effect of ELR-GNN with audio modal features is second, with average accuracy rates of 61.1% and 62.7% on the IEMOCAP and MELD data sets, and average F1 values of 60.8% and 62.0% respectively.",
            "[paragraph id = 5] ELR-GNN with video modality features has the worst emotion recognition effect, with average accuracy rates of 59.4% and 60.1% on the IEMOCAP and MELD data sets, and average F1 values of 59.7% and 61.4% respectively.",
            "[paragraph id = 6] Experimental results show that text features contain the most emotional semantic information.",
            "[paragraph id = 7] In the dual-modal experiment, ELR-GNN with text and audio modal features has the best emotion recognition effect.",
            "[paragraph id = 8] The average accuracy on the IEMOCAP and MELD data sets are 65.0% and 64.1%, respectively, and the average F1 values are are 64.4% and 63.2%, respectively.",
            "[paragraph id = 9] Experimental results demonstrate the effectiveness of multimodal features."
        ],
        "table_html": "<figure class=\"ltx_table\" id=\"S4.T4\">\n<figcaption class=\"ltx_caption\"><span class=\"ltx_tag ltx_tag_table\">Table 4: </span>The effect of ELR-GNN on IEMOCAP and MELD datasets using unimodal features and multimodal features, respectively. We report average accuracy and F1-score.</figcaption>\n<table class=\"ltx_tabular ltx_guessed_headers ltx_align_middle\" id=\"S4.T4.1\">\n<thead class=\"ltx_thead\">\n<tr class=\"ltx_tr\" id=\"S4.T4.1.1.1\">\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_th_row ltx_border_r ltx_border_t\" id=\"S4.T4.1.1.1.1\" rowspan=\"2\" style=\"padding:2.5pt 12.8pt;\"><span class=\"ltx_text\" id=\"S4.T4.1.1.1.1.1\">Modality</span></th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" colspan=\"2\" id=\"S4.T4.1.1.1.2\" style=\"padding:2.5pt 12.8pt;\">IEMOCAP</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" colspan=\"2\" id=\"S4.T4.1.1.1.3\" style=\"padding:2.5pt 12.8pt;\">MELD</th>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T4.1.2.2\">\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T4.1.2.2.1\" style=\"padding:2.5pt 12.8pt;\">Acc.</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T4.1.2.2.2\" style=\"padding:2.5pt 12.8pt;\">F1</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T4.1.2.2.3\" style=\"padding:2.5pt 12.8pt;\">Acc</th>\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_column ltx_border_t\" id=\"S4.T4.1.2.2.4\" style=\"padding:2.5pt 12.8pt;\">F1</th>\n</tr>\n</thead>\n<tbody class=\"ltx_tbody\">\n<tr class=\"ltx_tr\" id=\"S4.T4.1.3.1\">\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_row ltx_border_r ltx_border_t\" id=\"S4.T4.1.3.1.1\" style=\"padding:2.5pt 12.8pt;\">T</th>\n<td class=\"ltx_td ltx_align_center ltx_border_t\" id=\"S4.T4.1.3.1.2\" style=\"padding:2.5pt 12.8pt;\">64.1</td>\n<td class=\"ltx_td ltx_align_center ltx_border_t\" id=\"S4.T4.1.3.1.3\" style=\"padding:2.5pt 12.8pt;\">63.9</td>\n<td class=\"ltx_td ltx_align_center ltx_border_t\" id=\"S4.T4.1.3.1.4\" style=\"padding:2.5pt 12.8pt;\">63.5</td>\n<td class=\"ltx_td ltx_align_center ltx_border_t\" id=\"S4.T4.1.3.1.5\" style=\"padding:2.5pt 12.8pt;\">62.4</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T4.1.4.2\">\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_row ltx_border_r\" id=\"S4.T4.1.4.2.1\" style=\"padding:2.5pt 12.8pt;\">A</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T4.1.4.2.2\" style=\"padding:2.5pt 12.8pt;\">61.1</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T4.1.4.2.3\" style=\"padding:2.5pt 12.8pt;\">60.8</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T4.1.4.2.4\" style=\"padding:2.5pt 12.8pt;\">62.7</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T4.1.4.2.5\" style=\"padding:2.5pt 12.8pt;\">62.0</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T4.1.5.3\">\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_row ltx_border_r\" id=\"S4.T4.1.5.3.1\" style=\"padding:2.5pt 12.8pt;\">V</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T4.1.5.3.2\" style=\"padding:2.5pt 12.8pt;\">59.4</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T4.1.5.3.3\" style=\"padding:2.5pt 12.8pt;\">59.7</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T4.1.5.3.4\" style=\"padding:2.5pt 12.8pt;\">60.1</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T4.1.5.3.5\" style=\"padding:2.5pt 12.8pt;\">61.4</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T4.1.6.4\">\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_row ltx_border_r\" id=\"S4.T4.1.6.4.1\" style=\"padding:2.5pt 12.8pt;\">T+A</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T4.1.6.4.2\" style=\"padding:2.5pt 12.8pt;\">65.0</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T4.1.6.4.3\" style=\"padding:2.5pt 12.8pt;\">64.4</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T4.1.6.4.4\" style=\"padding:2.5pt 12.8pt;\">64.1</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T4.1.6.4.5\" style=\"padding:2.5pt 12.8pt;\">63.2</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T4.1.7.5\">\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_row ltx_border_r\" id=\"S4.T4.1.7.5.1\" style=\"padding:2.5pt 12.8pt;\">T+V</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T4.1.7.5.2\" style=\"padding:2.5pt 12.8pt;\">64.3</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T4.1.7.5.3\" style=\"padding:2.5pt 12.8pt;\">64.6</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T4.1.7.5.4\" style=\"padding:2.5pt 12.8pt;\">64.0</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T4.1.7.5.5\" style=\"padding:2.5pt 12.8pt;\">62.9</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T4.1.8.6\">\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_row ltx_border_r\" id=\"S4.T4.1.8.6.1\" style=\"padding:2.5pt 12.8pt;\">V+A</th>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T4.1.8.6.2\" style=\"padding:2.5pt 12.8pt;\">63.0</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T4.1.8.6.3\" style=\"padding:2.5pt 12.8pt;\">62.7</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T4.1.8.6.4\" style=\"padding:2.5pt 12.8pt;\">61.5</td>\n<td class=\"ltx_td ltx_align_center\" id=\"S4.T4.1.8.6.5\" style=\"padding:2.5pt 12.8pt;\">61.9</td>\n</tr>\n<tr class=\"ltx_tr\" id=\"S4.T4.1.9.7\">\n<th class=\"ltx_td ltx_align_center ltx_th ltx_th_row ltx_border_b ltx_border_r\" id=\"S4.T4.1.9.7.1\" style=\"padding:2.5pt 12.8pt;\">T+A+V</th>\n<td class=\"ltx_td ltx_align_center ltx_border_b\" id=\"S4.T4.1.9.7.2\" style=\"padding:2.5pt 12.8pt;\"><span class=\"ltx_text ltx_font_bold\" id=\"S4.T4.1.9.7.2.1\">70.6</span></td>\n<td class=\"ltx_td ltx_align_center ltx_border_b\" id=\"S4.T4.1.9.7.3\" style=\"padding:2.5pt 12.8pt;\"><span class=\"ltx_text ltx_font_bold\" id=\"S4.T4.1.9.7.3.1\">70.9</span></td>\n<td class=\"ltx_td ltx_align_center ltx_border_b\" id=\"S4.T4.1.9.7.4\" style=\"padding:2.5pt 12.8pt;\"><span class=\"ltx_text ltx_font_bold\" id=\"S4.T4.1.9.7.4.1\">68.7</span></td>\n<td class=\"ltx_td ltx_align_center ltx_border_b\" id=\"S4.T4.1.9.7.5\" style=\"padding:2.5pt 12.8pt;\"><span class=\"ltx_text ltx_font_bold\" id=\"S4.T4.1.9.7.5.1\">69.9</span></td>\n</tr>\n</tbody>\n</table>\n</figure>",
        "perturb_sentence_id": [
            3,
            4
        ],
        "output": {
            "perturbed_statement": "[paragraph id = 3] The average accuracy on the IEMOCAP and MELD data sets are 64.1% and 63.5%, respectively, and the average F1 value is 63.9% and 62.4%, respectively. The emotion recognition effect of ELR-GNN with audio modal features is second, with average accuracy rates of 61.1% and 62.7% on the IEMOCAP and MELD data sets, and average F1 values of 60.8% and 60.0% respectively.",
            "perturbed_explanation": "1. The original explanation provides the average accuracy and F1 values for ELR-GNN with audio modal features in the IEMOCAP and MELD datasets. It states that this configuration ranks second in terms of emotion recognition effect. 2. The statement alters the average F1 value for the MELD dataset from 62.0% to 60.0%. This change introduces a factual inaccuracy since the context specifies the original F1 value as 62.0%, not 60.0% as claimed."
        }
    }
]