4. 實驗結果
4.1. 實驗資料
本論文實驗目的為測試本論文的摘要擷取方法是否有更好的效能，本論文將 實驗果與邱怡菁(民 103)之論文所提到的摘要方法、曾士昌(民 99)之論文所提到 的摘要方法和線上摘要器 SweSum 進行比較以探討摘要擷取效能。為了探討本論 文所提出的文字稿摘要方法之效能，本論文從線上課程網站 Coursera 與 Edx 收 集三十篇課程文字稿與投影片，由於有些線上課程的某些章節的文字稿字詞數比 該章節的投影片字詞數還要來的少，因此本論文將實驗資料分為未刪除章節和刪 除章節兩部分。本論文除了用所有文字稿建立主題模型並擷取摘要外，本論文同 時使用文字稿將百分之八十的課程文字稿做為訓練資料集建立主題模型與百分 之二十的課程文字稿做為測試資料集擷取摘要進行比較。未刪除章節課程資訊如 表 2 所示，刪除章節課程資訊如表 3 所示，未刪除章節課程訓練資料集資訊如 表 4 所示，未刪除章節課程測試資料集資訊如表 5 所示，刪除章節課程訓練資 料集資訊如表 6 所示，刪除章節課程測試資料集資訊如表 7 所示，未刪除章節 課程的最低 Perplexity 值之主題數如表 8 所示，刪除章節課程的最低 Perplexity 值之主題數如表 9 所示，未刪除章節課程訓練資料集的最低 Perplexity 值之主題 數如表 10 所示，刪除章節課程訓練資料集的最低 Perplexity 值之主題數如表 11 所示。
表 2. 未刪除章節課程相關資訊
課程來源 課程名稱 章節
數
Chapter 數
句子 數
投影片 張數 coursera The Hardware/Software Interface 62 12 6004 621 coursera Social Network Analysis 26 8 2875 532 coursera Web Application Architectures 34 5 2238 194 coursera Audio Signal Processing for Music
Applications
22 9 2248 291
coursera Malicious Software and its Underground Economy Two Sides to
39 6 3332 383
22 Every Story
coursera Big Data Science with the BD2K-LINCS Data Coordination and Integration Center
28 12 1931 384
coursera Experimental Methods in Systems Biology
23 6 3244 286
coursera Dynamical Modeling Methods for Systems Biology
30 8 5737 342
coursera The Brain and Space 51 6 2635 669
coursera Network Analysis in Systems Biology
33 7 2440 395
coursera automata 23 7 5085 689
coursera Natural Language Processing 2013 116 20 6433 410 coursera Beginning Game Programming with
C#
49 8 4589 360
coursera Climate Change 22 4 2781 106
coursera Journalism Skills for Engaged Citizens
18 6 1457 115
coursera Algorithms Design and Analysis Part 1
74 10 10873 475
coursera Algorithms Design and Analysis Part 2
103 20 9991 572
coursera Introduction to Chemistry Reactions and Ratios
42 6 5250 446
coursera Genomic and Precision Medicine 31 7 3395 431 coursera Epigenetic Control of Gene
Expression
57 7 5166 281
coursera Take the Lead on Healthcare Quality Improvement
18 10 1885 446
coursera Surviving Disruptive Technologies 42 14 2997 425 coursera Caries Management by Risk
Assessment CAMBRA
51 6 5133 617
coursera Computational Neuroscience 28 8 4137 439 coursera Introduction to Data Science 93 10 7608 832
coursera Discrete Optimization 43 7 8540 546
coursera Foundations of Virtual Instruction 24 5 1569 133 coursera Virology I How Viruses Work 60 11 7246 605 edx Introduction to Computer Science 60 11 15177 2004
edx Big Data in Education 43 8 3970 1281
esx Cellular mechanisms of brain function
35 7 3936 471
edx Introduction to Programming with Java Part 1 Starting to Code with coursera The Hardware/Software Interface 61 12 5922 607
23
coursera Social Network Analysis 24 7 2670 462 coursera Web Application Architectures 33 5 2199 190 coursera Audio Signal Processing for Music
Applications
22 9 2248 291
coursera Malicious Software and its Underground Economy Two Sides to Every Story
37 6 3148 334
coursera Big Data Science with the BD2K-LINCS Data Coordination and Integration Center
20 8 1580 268
coursera Experimental Methods in Systems Biology
23 6 3244 286
coursera Dynamical Modeling Methods for Systems Biology
29 8 5533 308
coursera The Brain and Space 51 6 2635 669
coursera Network Analysis in Systems Biology
27 6 2114 322
coursera automata 23 7 5085 689
coursera Natural Language Processing 2013 115 20 6412 406 coursera Beginning Game Programming with
C#
49 8 4589 360
coursera Climate Change 22 4 2781 106
coursera Journalism Skills for Engaged Citizens
12 5 1037 73
coursera Algorithms Design and Analysis Part 1
74 10 10873 475
coursera Algorithms Design and Analysis Part 2
103 20 9991 572
coursera Introduction to Chemistry Reactions and Ratios
41 6 5134 433
coursera Genomic and Precision Medicine 20 7 2295 232 coursera Epigenetic Control of Gene
Expression
57 7 5166 281
coursera Take the Lead on Healthcare Quality Improvement
17 10 1838 417
coursera Surviving Disruptive Technologies 40 14 2908 403 coursera Caries Management by Risk
Assessment CAMBRA
51 6 5133 617
coursera Computational Neuroscience 28 8 4137 439 coursera Introduction to Data Science 86 10 7153 672
coursera Discrete Optimization 43 7 8540 546
coursera Foundations of Virtual Instruction 24 5 1569 133 coursera Virology I How Viruses Work 59 11 7052 591 edx Introduction to Computer Science 52 11 14671 1798
edx Big Data in Education 24 8 2167 615
esx Cellular mechanisms of brain function
35 7 3936 471
edx Introduction to Programming with Java Part 1 Starting to Code with Java
39 5 1599 222
edx Programming Basics 31 11 2506 421
24 coursera The Hardware/Software Interface 50 10 4654 487 coursera Social Network Analysis 21 6 2477 452 coursera Web Application Architectures 27 5 1810 149 coursera Audio Signal Processing for Music
Applications
18 8 1789 250
coursera Malicious Software and its Underground Economy Two Sides to Every Story
31 5 2590 294
coursera Big Data Science with the BD2K-LINCS Data Coordination and Integration Center
22 10 1393 318
coursera Experimental Methods in Systems Biology
19 5 2673 249
coursera Dynamical Modeling Methods for Systems Biology
24 6 4784 264
coursera The Brain and Space 41 5 2192 528
coursera Network Analysis in Systems Biology
27 6 2059 315
coursera automata 19 5 4166 565
coursera Natural Language Processing 2013 93 16 5382 341 coursera Beginning Game Programming with
C#
40 6 3927 303
coursera Climate Change 18 3 2281 84
coursera Journalism Skills for Engaged Citizens
15 4 1219 96
coursera Algorithms Design and Analysis Part 1
60 10 8599 379
coursera Algorithms Design and Analysis Part 2
83 17 8105 448
coursera Introduction to Chemistry Reactions and Ratios
34 5 4361 357
coursera Genomic and Precision Medicine 25 6 2637 331 coursera Epigenetic Control of Gene
Expression
46 6 4007 230
coursera Take the Lead on Healthcare Quality Improvement
15 8 1575 381
coursera Surviving Disruptive Technologies 34 12 2508 350 coursera Caries Management by Risk
Assessment CAMBRA
41 5 4035 492
coursera Computational Neuroscience 23 7 3423 405 coursera Introduction to Data Science 75 8 6510 697
coursera Discrete Optimization 35 5 6724 442
coursera Foundations of Virtual Instruction 20 4 1452 119 coursera Virology I How Viruses Work 48 10 5980 509
25 edx Introduction to Programming with
Java Part 1 Starting to Code with coursera The Hardware/Software Interface 12 3 1350 134
coursera Social Network Analysis 5 3 398 80
coursera Web Application Architectures 7 1 428 45 coursera Audio Signal Processing for Music
Applications
4 2 459 41
coursera Malicious Software and its Underground Economy Two Sides to Every Story
8 2 742 89
coursera Big Data Science with the BD2K-LINCS Data Coordination and Integration Center
6 2 538 66
coursera Experimental Methods in Systems Biology
4 2 571 37
coursera Dynamical Modeling Methods for Systems Biology
6 3 953 78
coursera The Brain and Space 10 2 443 141
coursera Network Analysis in Systems Biology
6 2 381 80
coursera automata 4 2 919 124
coursera Natural Language Processing 2013 23 4 1051 69 coursera Beginning Game Programming with
C#
9 3 662 57
coursera Climate Change 4 1 500 22
coursera Journalism Skills for Engaged Citizens
3 3 238 19
coursera Algorithms Design and Analysis Part 1
14 4 2274 96
coursera Algorithms Design and Analysis Part 2
20 4 1886 124
coursera Introduction to Chemistry Reactions and Ratios
8 2 889 89
coursera Genomic and Precision Medicine 6 2 758 100 coursera Epigenetic Control of Gene
Expression
11 1 1159 51
coursera Take the Lead on Healthcare Quality Improvement
3 3 310 65
coursera Surviving Disruptive Technologies 8 3 489 75 coursera Caries Management by Risk
Assessment CAMBRA
10 1 1098 125
coursera Computational Neuroscience 5 2 714 34 coursera Introduction to Data Science 18 3 1098 135
coursera Discrete Optimization 8 3 1816 104
26
coursera Foundations of Virtual Instruction 4 1 117 14 coursera Virology I How Viruses Work 12 2 1266 96
edx Introduction to Programming with Java Part 1 Starting to Code with coursera The Hardware/Software Interface 49 10 4572 473 coursera Social Network Analysis 20 6 2346 415 coursera Web Application Architectures 27 5 1737 157 coursera Audio Signal Processing for Music
Applications
18 8 1789 250
coursera Malicious Software and its Underground Economy Two Sides to Every Story
30 5 2494 274
coursera Big Data Science with the BD2K-LINCS Data Coordination and Integration Center
16 7 1327 231
coursera Experimental Methods in Systems Biology
19 5 2673 249
coursera Dynamical Modeling Methods for Systems Biology
24 6 4866 256
coursera The Brain and Space 41 5 2192 528
coursera Network Analysis in Systems Biology
22 4 1759 259
coursera automata 19 5 4166 565
coursera Natural Language Processing 2013 93 17 5366 338 coursera Beginning Game Programming with
C#
40 6 3927 303
coursera Climate Change 18 3 2281 84
coursera Journalism Skills for Engaged Citizens
10 4 836 59
coursera Algorithms Design and Analysis Part 1
60 10 8599 379
coursera Algorithms Design and Analysis Part 2
83 17 8105 448
coursera Introduction to Chemistry Reactions and Ratios
33 5 4245 344
coursera Genomic and Precision Medicine 16 6 1797 174 coursera Epigenetic Control of Gene
Expression
46 6 4007 230
coursera Take the Lead on Healthcare Quality Improvement
14 8 1528 352
coursera Surviving Disruptive Technologies 32 12 2419 328
27 coursera Caries Management by Risk
Assessment CAMBRA
41 5 4035 492
coursera Computational Neuroscience 23 7 3423 405 coursera Introduction to Data Science 69 8 6051 543
coursera Discrete Optimization 35 5 6724 442
coursera Foundations of Virtual Instruction 20 4 1452 119 coursera Virology I How Viruses Work 48 10 5932 506
edx Introduction to Programming with Java Part 1 Starting to Code with coursera The Hardware/Software Interface 12 3 1350 134
coursera Social Network Analysis 4 2 324 47
coursera Web Application Architectures 6 1 462 33 coursera Audio Signal Processing for Music
Applications
4 2 459 41
coursera Malicious Software and its Underground Economy Two Sides to Every Story
7 2 654 60
coursera Big Data Science with the BD2K-LINCS Data Coordination and Integration Center
4 2 253 37
coursera Experimental Methods in Systems Biology
4 2 571 37
coursera Dynamical Modeling Methods for Systems Biology
5 2 667 52
coursera The Brain and Space 10 2 443 141
coursera Network Analysis in Systems Biology
5 2 355 63
coursera automata 4 2 919 124
coursera Natural Language Processing 2013 22 4 1046 68 coursera Beginning Game Programming with
C#
9 3 662 57
coursera Climate Change 4 1 500 22
coursera Journalism Skills for Engaged Citizens
2 1 201 14
coursera Algorithms Design and Analysis Part 1
14 4 2274 96
coursera Algorithms Design and Analysis Part 2
20 4 1886 124
coursera Introduction to Chemistry Reactions and Ratios
8 2 889 89
28
coursera Genomic and Precision Medicine 4 1 498 58 coursera Epigenetic Control of Gene
Expression
11 1 1159 51
coursera Take the Lead on Healthcare Quality Improvement
3 3 310 65
coursera Surviving Disruptive Technologies 8 3 489 75 coursera Caries Management by Risk
Assessment CAMBRA
10 1 1098 125
coursera Computational Neuroscience 5 2 714 34 coursera Introduction to Data Science 17 3 1102 129
coursera Discrete Optimization 8 3 1816 104
coursera Foundations of Virtual Instruction 4 1 117 14 coursera Virology I How Viruses Work 11 2 1120 85
edx Introduction to Programming with Java Part 1 Starting to Code with
The Hardware/Software Interface 89
Social Network Analysis 51
Web Application Architectures 51
Audio Signal Processing for Music Applications 49 Malicious Software and its Underground Economy Two Sides to
Experimental Methods in Systems Biology 70
Dynamical Modeling Methods for Systems Biology 73
The Brain and Space 85
Network Analysis in Systems Biology 66
automata 40
Natural Language Processing 2013 37
Beginning Game Programming with C# 13
Climate Change 51
Journalism Skills for Engaged Citizens 40
Algorithms Design and Analysis Part 1 43
Algorithms Design and Analysis Part 2 48
Introduction to Chemistry Reactions and Ratios 19
Genomic and Precision Medicine 59
Epigenetic Control of Gene Expression 92
Take the Lead on Healthcare Quality Improvement 42
Surviving Disruptive Technologies 1
Caries Management by Risk Assessment CAMBRA 10
29
Computational Neuroscience 59
Introduction to Data Science 12
Discrete Optimization 25
Foundations of Virtual Instruction 16
Virology I How Viruses Work 29
Edx Introduction to Computer Science 36
Edx Big Data in Education 165
Edx Cellular mechanisms of brain function 10
Edx Introduction to Programming with Java Part 1 Starting to Code with Java
44
Edx Programming Basics 91
表 9. 刪章節時各文字稿課程最低 Perplexity 值之主題數
課程名稱 最低 Perplexity 值
之主題數
The Hardware/Software Interface 99
Social Network Analysis 55
Web Application Architectures 62
Audio Signal Processing for Music Applications 49 Malicious Software and its Underground Economy Two Sides to
Experimental Methods in Systems Biology 70
Dynamical Modeling Methods for Systems Biology 73
The Brain and Space 85
Network Analysis in Systems Biology 55
automata 40
Natural Language Processing 2013 32
Beginning Game Programming with C# 13
Climate Change 51
Journalism Skills for Engaged Citizens 40
Algorithms Design and Analysis Part 1 43
Algorithms Design and Analysis Part 2 48
Introduction to Chemistry Reactions and Ratios 18
Genomic and Precision Medicine 50
Epigenetic Control of Gene Expression 92
Take the Lead on Healthcare Quality Improvement 39
Surviving Disruptive Technologies 71
Caries Management by Risk Assessment CAMBRA 10
Computational Neuroscience 59
Introduction to Data Science 122
Discrete Optimization 25
Foundations of Virtual Instruction 16
Virology I How Viruses Work 94
Edx Introduction to Computer Science 23
Edx Big Data in Education 187
Edx Cellular mechanisms of brain function 10
Edx Introduction to Programming with Java Part 1 Starting to Code with Java
10
30
Edx Programming Basics 55
表 10. 未刪章節時訓練資料集最低 Perplexity 值之主題數
課程名稱 最低 Perplexity 值
之主題數
The Hardware/Software Interface 38
Social Network Analysis 145
Web Application Architectures 76
Audio Signal Processing for Music Applications 139 Malicious Software and its Underground Economy Two Sides to
Every Story
35 Big Data Science with the BD2K-LINCS Data Coordination and
Integration Center
38
Experimental Methods in Systems Biology 144
Dynamical Modeling Methods for Systems Biology 54
The Brain and Space 79
Network Analysis in Systems Biology 12
automata 52
Natural Language Processing 2013 37
Beginning Game Programming with C# 26
Climate Change 42
Journalism Skills for Engaged Citizens 1
Algorithms Design and Analysis Part 1 32
Algorithms Design and Analysis Part 2 9
Introduction to Chemistry Reactions and Ratios 58
Genomic and Precision Medicine 18
Epigenetic Control of Gene Expression 19
Take the Lead on Healthcare Quality Improvement 1
Surviving Disruptive Technologies 71
Caries Management by Risk Assessment CAMBRA 14
Computational Neuroscience 156
Introduction to Data Science 125
Discrete Optimization 17
Foundations of Virtual Instruction 53
Virology I How Viruses Work 1
Edx Introduction to Computer Science 34
Edx Big Data in Education 70
Edx Cellular mechanisms of brain function 68
Edx Introduction to Programming with Java Part 1 Starting to Code with Java
91
Edx Programming Basics 39
表 11.
刪章節時訓練資料集最低 Perplexity 值之主題數
課程名稱 最低 Perplexity 值
之主題數
The Hardware/Software Interface 1
Social Network Analysis 121
Web Application Architectures 17
31
Audio Signal Processing for Music Applications 139 Malicious Software and its Underground Economy Two Sides to
Every Story
74 Big Data Science with the BD2K-LINCS Data Coordination and
Integration Center
1
Experimental Methods in Systems Biology 144
Dynamical Modeling Methods for Systems Biology 71
The Brain and Space 79
Network Analysis in Systems Biology 61
automata 52
Natural Language Processing 2013 95
Beginning Game Programming with C# 26
Climate Change 42
Journalism Skills for Engaged Citizens 1
Algorithms Design and Analysis Part 1 32
Algorithms Design and Analysis Part 2 9
Introduction to Chemistry Reactions and Ratios 1
Genomic and Precision Medicine 109
Epigenetic Control of Gene Expression 19
Take the Lead on Healthcare Quality Improvement 1
Surviving Disruptive Technologies 174
Caries Management by Risk Assessment CAMBRA 14
Computational Neuroscience 156
Introduction to Data Science 70
Discrete Optimization 17
Foundations of Virtual Instruction 53
Virology I How Viruses Work 13
Edx Introduction to Computer Science 55
Edx Big Data in Education 1
Edx Cellular mechanisms of brain function 68
Edx Introduction to Programming with Java Part 1 Starting to Code with Java
1
Edx Programming Basics 1