[ { "id": 0, "page": 6, "bounding_box": [ 72.0, 517.4150187174479, 296.14898681640625, 600.9990234375 ], "latex_content": "\\begin{table}[h]\n\\centering\n\\begin{tabular}{|c|c|c|c|c|}\n\\hline\n & \\multicolumn{2}{c|}{CNN} & \\multicolumn{2}{c|}{Daily Mail} \\\\ \\hline\n & Train & Test & Train & Test \\\\ \\hline\nDeep LSTM$^\\dagger$ & 55 & 57 & 63.3 & 62.2 \\\\ \\hline\nAttentive Reader$^\\dagger$ & 61.6 & 63 & \\textbf{70.5} & \\textbf{69} \\\\ \\hline\nImpatient Reader$^\\dagger$ & 61.8 & 63.8 & 69 & 68 \\\\ \\hline\nL2R Reader & \\textbf{64.3} & \\textbf{65.8} & 69.1 & 67.3 \\\\ \\hline\n\\end{tabular}\n\\label{result}\n\\caption{Results of our L2R Reader on the CNN and Daily Mail dataset. Results marked with $^\\dagger$ are taken from previous paper.}\n\\end{table}", "extracted_content": [ [ "", "CNN", null, "Daily Mail", null ], [ "", "Train", "Test", "Train", "Test" ], [ "Deep LSTM\u2020", "55", "57", "63.3", "62.2" ], [ "Attentive Reader\u2020", "61.6", "63", "70.5", "69" ], [ "Impatient Reader\u2020", "61.8", "63.8", "69", "68" ], [ "L2R Reader", "64.3", "65.8", "69.1", "67.3" ] ], "similarity_score": 0.6611226611226612, "table_image": "images/1605.03284v2/table_0.png", "page_image": "pages/1605.03284v2/page_6.png" }, { "id": 1, "page": 6, "bounding_box": [ 340.0610046386719, 114.885009765625, 513.1390713778409, 295.010009765625 ], "latex_content": "\\begin{table}[h]\n\\centering\n\\begin{tabular}{|c|c|c|}\n\\hline\n\\begin{tabular}[c]{@{}c@{}}Ranking\\\\ Model\\end{tabular} & \\begin{tabular}[c]{@{}c@{}}CNN\\\\ Test\\end{tabular} & \\begin{tabular}[c]{@{}c@{}}Daily Mail \\\\ Test\\end{tabular} \\\\ \\hline\nRankSVM & \\textbf{65.8} & 66.7 \\\\ \\hline\nMART & 60.4 & 65.3 \\\\ \\hline\nRankNet & 40.9 & 32.8 \\\\ \\hline\nRankBoost & 32.0 & 28.4 \\\\ \\hline\nAdaRank & 18.0 & 12.7 \\\\ \\hline\n\\begin{tabular}[c]{@{}c@{}}Coordinate \\\\ Asecent\\end{tabular} & 59.0 & 54.4 \\\\ \\hline\nLambdaMART & 64.2 & \\textbf{67.3} \\\\ \\hline\nListNet & 32.7 & 32.3 \\\\ \\hline\n\\begin{tabular}[c]{@{}c@{}}Random \\\\ Forest\\end{tabular} & 63.4 & 65.6 \\\\ \\hline\n\\end{tabular}\n\\caption{Performance of different L2R algorithms on CNN and Daily Mail dataset.}\n\\label{L2R}\n\\end{table}", "extracted_content": [ [ "Ranking\nModel", "CNN\nTest", "Daily Mail\nTest" ], [ "RankSVM", "65.8", "66.7" ], [ "MART", "60.4", "65.3" ], [ "RankNet", "40.9", "32.8" ], [ "RankBoost", "32.0", "28.4" ], [ "AdaRank", "18.0", "12.7" ], [ "Coordinate\nAsecent", "59.0", "54.4" ], [ "LambdaMART", "64.2", "67.3" ], [ "ListNet", "32.7", "32.3" ], [ "Random\nForest", "63.4", "65.6" ] ], "similarity_score": 0.6128266033254157, "table_image": "images/1605.03284v2/table_1.png", "page_image": "pages/1605.03284v2/page_6.png" }, { "id": 2, "page": 6, "bounding_box": [ 331.4320068359375, 433.989013671875, 521.7670288085938, 531.4923502604166 ], "latex_content": "\\begin{table}[h]\n\\centering\n\\begin{tabular}{|l|l|l|l|l|}\n\\hline\n & \\multicolumn{2}{l|}{CNN} & \\multicolumn{2}{l|}{Daily Mail} \\\\ \\hline\n & Train & Test & Train & Test \\\\ \\hline\nFrequency & 33.2 & 35.0 & 33.4 & 32.4 \\\\ \\hline\nWA & 47.3 & 50.6 & 54.3 & 53.4 \\\\ \\hline\nnBOW & 40.3 & 43.5 & 48.7 & 47.8 \\\\ \\hline\nWMD & 41.2 & 44.0 & 49.3 & 48.5 \\\\ \\hline\nFS & 18.9 & 22.0 & 25.2 & 24.3 \\\\ \\hline\n\\end{tabular}\n\\caption{Single feature performance on CNN and Daily Mail dataset.}\n\\label{single feature}\n\\end{table}", "extracted_content": [ [ "", "CNN", null, "Daily Mail", null ], [ "", "Train", "Test", "Train", "Test" ], [ "Frequency", "33.2", "35.0", "33.4", "32.4" ], [ "WA", "47.3", "50.6", "54.3", "53.4" ], [ "nBOW", "40.3", "43.5", "48.7", "47.8" ], [ "WMD", "41.2", "44.0", "49.3", "48.5" ], [ "FS", "18.9", "22.0", "25.2", "24.3" ] ], "similarity_score": 0.7834549878345499, "table_image": "images/1605.03284v2/table_2.png", "page_image": "pages/1605.03284v2/page_6.png" }, { "id": 3, "page": 7, "bounding_box": [ 96.00900268554688, 72.198974609375, 274.79105704171315, 253.51898193359375 ], "latex_content": "\\begin{table}[h]\n\\centering\n\\begin{tabular}{|c|c|c|c|c|}\n\\hline\n & \\multicolumn{2}{l|}{CNN} & \\multicolumn{2}{l|}{Dailymail} \\\\ \\hline\n\\#Training & Val & Test & Val & Test \\\\ \\hline\n10 & 35.2 & 41.6 & 45.7 & 43.6 \\\\ \\hline\n20 & 57.4 & 55.1 & 57.2 & 56.2 \\\\ \\hline\n30 & 56.7 & 60.2 & 61.4 & 60.3 \\\\ \\hline\n40 & 57 & 60 & 62.3 & 61.3 \\\\ \\hline\n50 & 60.3 & 63.1 & 63.5 & 62.5 \\\\ \\hline\n100 & 61.5 & 63.9 & 65.4 & 64.6 \\\\ \\hline\n200 & 62.5 & 64.9 & 67.3 & 65.2 \\\\ \\hline\n500 & 62.8 & 65 & 67.5 & 66.3 \\\\ \\hline\n1000 & 62.9 & 65.2 & 68.3 & 66.7 \\\\ \\hline\n2000 & 63.2 & 65.2 & 69.0 & 67.3 \\\\ \\hline\n5000 & 64.3 & 65.8 & 69.1 & 67.3 \\\\ \\hline\n\\end{tabular}\n\\caption{Model performance for L2R Reader against different number of training data on the CNN and Daily Mail dataset. All the statistics are averaged over 10 times of random samples. Parameters are tuned on the validation set and the best validation model is applied to the test data.}\n\\label{performance vs data size table}\n\\end{table}", "extracted_content": [ [ "", "CNN", null, "Dailymail", null ], [ "#Training", "Val", "Test", "Val", "Test" ], [ "10", "35.2", "41.6", "45.7", "43.6" ], [ "20", "57.4", "55.1", "57.2", "56.2" ], [ "30", "56.7", "60.2", "61.4", "60.3" ], [ "40", "57", "60", "62.3", "61.3" ], [ "50", "60.3", "63.1", "63.5", "62.5" ], [ "100", "61.5", "63.9", "65.4", "64.6" ], [ "200", "62.5", "64.9", "67.3", "65.2" ], [ "500", "62.8", "65", "67.5", "66.3" ], [ "1000", "62.9", "65.2", "68.3", "66.7" ], [ "2000", "63.2", "65.2", "69.0", "67.3" ], [ "5000", "64.3", "65.8", "69.1", "67.3" ] ], "similarity_score": 0.6470588235294118, "table_image": "images/1605.03284v2/table_3.png", "page_image": "pages/1605.03284v2/page_7.png" }, { "id": 4, "page": 7, "bounding_box": [ 90.8219985961914, 412.1650085449219, 279.97900390625, 495.85198974609375 ], "latex_content": "\\begin{table}[h]\n\\centering\n\\begin{tabular}{|c|c|c|c|c|}\n\\hline\n & \\multicolumn{2}{c|}{CNN} & \\multicolumn{2}{c|}{Daily Mail} \\\\ \\hline\nModels & Val & Test & Val & Test \\\\ \\hline\nL2R Reader & 64.3 & 65.8 & 69.1 & 67.3 \\\\ \\hline\nL2R+Coref & 63.8 & 64.8 & 68.3 & 66.5 \\\\ \\hline\nL2R-WMD & 60.8 & 61.5 & 63.2 & 61.6 \\\\ \\hline\nL2R-FS & 61.5 & 62.5 & 65.3 & 63.7 \\\\ \\hline\n\\end{tabular}\n\\caption{Analysis of semantic components of our model. ``+\"\" and ``-\" refer to added or ablated components. Coref, WMD and FS denotes coreference system, word mover's distance and frame semantics.}\n\\label{semantics}\n\\end{table}", "extracted_content": [ [ "", "CNN", null, "Daily Mail", null ], [ "Models", "Val", "Test", "Val", "Test" ], [ "L2R Reader", "64.3", "65.8", "69.1", "67.3" ], [ "L2R+Coref", "63.8", "64.8", "68.3", "66.5" ], [ "L2R-WMD", "60.8", "61.5", "63.2", "61.6" ], [ "L2R-FS", "61.5", "62.5", "65.3", "63.7" ] ], "similarity_score": 0.5916030534351145, "table_image": "images/1605.03284v2/table_4.png", "page_image": "pages/1605.03284v2/page_7.png" } ]