7. 附錄
7.4. 測試資料集以章節數為主題數
未刪章節未加入提示詞時與一般 LDA 摘要方法和 PLSA 摘要方法比較結果 如表 82 所示,與動詞 LDA 摘要方法和名詞 LDA 摘要方法比較結果如表 83 所 示,與線上摘要器 SweSum 比較結果如表 84 所示。
表 82. 與一般 LDA 摘要方法和 PLSA 摘要方法比較結果
課程名稱 一般
LDA
TF TF-ISF TF-SF PLSA
The Hardware/Software Interface
0.3747 0.3783 0.3807 0.3753 0.3655 Social Network Analysis 0.3334 0.328 0.3296 0.3128 0.3072 Web Application Architectures 0.4834 0.4958 0.493 0.492 0.4785 Audio Signal Processing for
Music Applications
0.2146 0.2349 0.226 0.2399 0.2033 Malicious Software and its
Underground Economy Two Sides to Every Story
0.4268 0.4418 0.4384 0.4479 0.4275
Big Data Science with the
BD2K-LINCS Data
Coordination and Integration Center
0.2789 0.3273 0.3155 0.3119 0.3132
Experimental Methods in Systems Biology
0.0883 0.0986 0.1003 0.099 0.094 Dynamical Modeling Methods
for Systems Biology
0.3757 0.3969 0.3942 0.3893 0.383 The Brain and Space 0.3288 0.3112 0.3139 0.3104 0.3163 Network Analysis in Systems
Biology
0.4053 0.4113 0.4099 0.4036 0.3914
automata 0.5575 0.5637 0.5617 0.5602 0.5384
Natural Language Processing 2013
0.2466 0.2563 0.246 0.2642 0.2644
Beginning Game
Programming with C
0.233 0.2841 0.2705 0.2769 0.2213 Climate Change 0.221 0.2145 0.212 0.2121 0.2262 Journalism Skills for Engaged
Citizens
0.4183 0.4383 0.4285 0.4445 0.4136 Algorithms Design and 0.3293 0.3504 0.3441 0.3555 0.3105
174 Analysis Part 1
Algorithms Design and Analysis Part 2
0.3405 0.3677 0.3617 0.3621 0.3465 Introduction to Chemistry
Reactions and Ratios
0.217 0.2119 0.2195 0.2133 0.2215 Genomic and Precision
Medicine Surviving Disruptive
Technologies
0.4119 0.3936 0.3954 0.3798 0.3907 Caries Management by Risk
Assessment CAMBRA
0.351 0.4488 0.4347 0.4207 0.3233 Computational Neuroscience 0.3232 0.3144 0.3125 0.3101 0.3086 Introduction to Data Science 0.2965 0.3035 0.3008 0.3095 0.2996 Discrete Optimization 0.2971 0.3056 0.3019 0.3067 0.2914 Foundations of Virtual
Instruction
0.1972 0.2803 0.2807 0.2823 0.2919 Virology I How Viruses Work 0.2312 0.2524 0.2447 0.2454 0.221 Edx Introduction to Computer
Science Starting to Code with Java
0.1464 0.14 0.1371 0.1413 0.1315
edx Programming Basics 0.3981 0.3874 0.391 0.3812 0.3943 總平均 0.314 0.3297 0.3266 0.3268 0.3145
平均 F1 值:與一般 LDA 摘要方法呈現顯著差異
平均 F1 值:與 PLSA 摘要方法呈現顯著差異
紅色字體平均 F1 值:平均 F1 值未比一般 LDA 摘要方法或 PLSA 摘要方法高
在表 82 中除了 Social Network Analysis 課程、The Brain and Space 課程、
automata 課程、Climate Change 課程、Introduction to Chemistry Reactions and Ratios
課程、Genomic and Precision Medicine 課程、Surviving Disruptive Technologies 課
程、Computational Neuroscience 課程、Foundations of Virtual Instruction 課程、edx
Cellular mechanisms of brain function 課程、edx Introduction to Programming with
Java Part 1 Starting to Code with Java 課程和 edx Programming Basics 課程的 TF 摘
要方法、TF-ISF 摘要方法和 TF 摘要方法平均 F1 值皆比一般 LDA 摘要方法或
PLSA 摘要方法低,其餘課程的 TF 摘要方法、TF-ISF 摘要方法和 TF-SF 摘要方
175 The Hardware/Software
Interface
0.3785 0.3783 0.3807 0.3753 0.3693 Social Network Analysis 0.3238 0.328 0.3296 0.3128 0.3221 Web Application Architectures 0.4833 0.4958 0.493 0.492 0.4938 Audio Signal Processing for
Music Applications
0.2191 0.2349 0.226 0.2399 0.2212 Malicious Software and its
Underground Economy Two Sides to Every Story
0.4251 0.4418 0.4384 0.4479 0.4411
Big Data Science with the
BD2K-LINCS Data
Coordination and Integration Center
0.2981 0.3273 0.3155 0.3119 0.3084
Experimental Methods in Systems Biology
0.1081 0.0986 0.1003 0.099 0.0865 Dynamical Modeling Methods
for Systems Biology
0.3823 0.3969 0.3942 0.3893 0.3634 The Brain and Space 0.3019 0.3112 0.3139 0.3104 0.3337 Network Analysis in Systems
Biology
0.3885 0.4113 0.4099 0.4036 0.3848
automata 0.5445 0.5637 0.5617 0.5602 0.559
Natural Language Processing 2013
0.2449 0.2563 0.246 0.2642 0.2521 Beginning Game Programming
with C
0.2425 0.2841 0.2705 0.2769 0.2711 Climate Change 0.2085 0.2145 0.212 0.2121 0.2253 Journalism Skills for Engaged
Citizens
0.4352 0.4383 0.4285 0.4445 0.4249 Algorithms Design and Analysis
Part 1
0.3123 0.3504 0.3441 0.3555 0.2956 Algorithms Design and Analysis
Part 2
0.3583 0.3677 0.3617 0.3621 0.3475 Introduction to Chemistry
Reactions and Ratios
0.2447 0.2119 0.2195 0.2133 0.2274 Genomic and Precision
Medicine Surviving Disruptive
Technologies
0.3743 0.3936 0.3954 0.3798 0.3847 Caries Management by Risk
Assessment CAMBRA
0.3921 0.4488 0.4347 0.4207 0.3702
176
Computational Neuroscience 0.3069 0.3144 0.3125 0.3101 0.2996 Introduction to Data Science 0.3067 0.3035 0.3008 0.3095 0.3073 Discrete Optimization 0.2824 0.3056 0.3019 0.3067 0.2928 Foundations of Virtual
Instruction
0.1633 0.2803 0.2807 0.2823 0.2472 Virology I How Viruses Work 0.2322 0.2524 0.2447 0.2454 0.2286 Edx Introduction to Computer
Science
0.2591 0.2717 0.2736 0.2738 0.2652 edx Big Data in Education 0.4921 0.493 0.5014 0.4869 0.4854 edx Cellular mechanisms of
brain function
0.1744 0.1716 0.1697 0.178 0.1794 edx Introduction to
Programming with Java Part 1 Starting to Code with Java
0.1276 0.14 0.1371 0.1413 0.1495
edx Programming Basics 0.3932 0.3874 0.391 0.3812 0.3927 總平均 0.3137 0.3297 0.3266 0.3268 0.318
平均 F1 值:與動詞 LDA 摘要方法呈現顯著差異
平均 F1 值:與名詞 LDA 摘要方法呈現顯著差異
紅色字體平均 F1 值:平均 F1 值未比動詞 LDA 摘要方法或名詞 LDA 摘要方法高
在表 83 的結果中除了 Experimental Methods in Systems Biology 課程、The Brain and Space 課程、Climate Change 課程、Introduction to Chemistry Reactions and Ratios 課程、Genomic and Precision Medicine 課程、Computational Neuroscience 課 程 、 edx Cellular mechanisms of brain function 課 程 、 edx Introduction to Programming with Java Part 1 Starting to Code with Java 課程和 edx Programming Basics 課程的 TF 摘要方法、TF-ISF 摘要方法和 TF-SF 摘要方法平均 F1 值皆比 動詞 LDA 摘要方法和名詞 LDA 摘要方法低,其餘課程的 TF 摘要方法、TF-ISF 摘要方法和 TF-SF 摘要方法平均 F1 值至少有一個摘要方法平均 F1 值比動詞 LDA 摘要方法和名詞 LDA 摘要方法高。
表 84. 與線上摘要器 SweSum 比較結果
課程名稱 TF TF-ISF TF-SF SweSum
The Hardware/Software Interface 0.3783 0.3807 0.3753 0.3797 Social Network Analysis 0.328 0.3296 0.3128 0.3371 Web Application Architectures 0.4958 0.493 0.492 0.5109 Audio Signal Processing for Music
Applications
0.2349 0.226 0.2399 0.2495 Malicious Software and its Underground
Economy Two Sides to Every Story
0.4418 0.4384 0.4479 0.439
177 Big Data Science with the BD2K-LINCS Data Coordination and Integration Center
0.3273 0.3155 0.3119 0.2951 Experimental Methods in Systems Biology 0.0986 0.1003 0.099 0.091 Dynamical Modeling Methods for Systems
Biology
0.3969 0.3942 0.3893 0.3794 The Brain and Space 0.3112 0.3139 0.3104 0.3422 Network Analysis in Systems Biology 0.4113 0.4099 0.4036 0.3993
automata 0.5637 0.5617 0.5602 0.5394
Natural Language Processing 2013 0.2563 0.246 0.2642 0.2429 Beginning Game Programming with C 0.2841 0.2705 0.2769 0.2662
Climate Change 0.2145 0.212 0.2121 0.2223
Journalism Skills for Engaged Citizens 0.4383 0.4285 0.4445 0.4255 Algorithms Design and Analysis Part 1 0.3504 0.3441 0.3555 0.3238 Algorithms Design and Analysis Part 2 0.3677 0.3617 0.3621 0.3606 Introduction to Chemistry Reactions and
Ratios
0.2119 0.2195 0.2133 0.2034 Genomic and Precision Medicine 0.2387 0.2365 0.2362 0.2358 Epigenetic Control of Gene Expression 0.3404 0.3271 0.3362 0.3349 Take the Lead on Healthcare Quality
Improvement
0.4261 0.4251 0.4259 0.3841 Surviving Disruptive Technologies 0.3936 0.3954 0.3798 0.3947 Caries Management by Risk Assessment
CAMBRA
0.4488 0.4347 0.4207 0.4158 Computational Neuroscience 0.3144 0.3125 0.3101 0.3261 Introduction to Data Science 0.3035 0.3008 0.3095 0.3121 Discrete Optimization 0.3056 0.3019 0.3067 0.2944 Foundations of Virtual Instruction 0.2803 0.2807 0.2823 0.2364 Virology I How Viruses Work 0.2524 0.2447 0.2454 0.2431 Edx Introduction to Computer Science 0.2717 0.2736 0.2738 0.249 edx Big Data in Education 0.493 0.5014 0.4869 0.506 edx Cellular mechanisms of brain function 0.1716 0.1697 0.178 0.1721 edx Introduction to Programming with Java
Part 1 Starting to Code with Java
0.14 0.1371 0.1413 0.1426 edx Programming Basics 0.3874 0.391 0.3812 0.3733
總平均 0.3297 0.3266 0.3268 0.3221
平均 F1 值:與線上摘要器 SweSum 呈現顯著差異
紅色字體平均 F1 值:平均 F1 值未比線上摘要器 SweSum 高
在表 84 的結果中除了 Social Network Analysis 課程、Web Application
Architectures 課 程 、 Audio Signal Processing for Music Applications 課 程 、
Experimental Methods in Systems Biology 課程、The Brain and Space 課程、Climate
Change 課程、Genomic and Precision Medicine 課程、Take the Lead on Healthcare
Quality Improvement 課 程 、 Computational Neuroscience 課 程 、 edx Cellular
mechanisms of brain function 課程、edx Introduction to Programming with Java Part
178
1 Starting to Code with Java 課程和 edx Programming Basics 課程的 TF 摘要方法、
TF-ISF 摘要方法和 TF-SF 摘要方法平均 F1 值皆比線上摘要器 SweSum 低,其餘 課程的 TF 摘要方法、TF-ISF 摘要方法和 TF-SF 摘要方法平均 F1 值至少有一個 摘要方法比線上摘要器 SweSum 高。
未刪章節使用提示詞 TF 權重加權提示詞時與一般 LDA 摘要方法和 PLSA 摘要方法比較結果如表 85 所示,與動詞 LDA 摘要方法和名詞 LDA 摘要方法比 較結果如表 86 所示,與線上摘要器 SweSum 比較結果如表 87 所示。
表 85. 與一般 LDA 摘要方法和 PLSA 摘要方法比較結果
課程名稱 一般
LDA
TF TF-ISF TF-SF PLSA The Hardware/Software Interface 0.3747 0.3707 0.3763 0.3619 0.3655 Social Network Analysis 0.3334 0.3168 0.3168 0.3106 0.3072 Web Application Architectures 0.4834 0.4881 0.4885 0.4682 0.4785 Audio Signal Processing for Music
Applications
0.2146 0.225 0.2286 0.2096 0.2033 Malicious Software and its Underground
Economy Two Sides to Every Story
0.4268 0.4167 0.4274 0.4129 0.4275 Big Data Science with the
BD2K-LINCS Data Coordination and Integration Center
0.2789 0.3163 0.314 0.2953 0.3132
Experimental Methods in Systems Biology
0.0883 0.0965 0.1016 0.0868 0.094 Dynamical Modeling Methods for
Systems Biology
0.3757 0.3847 0.39 0.3675 0.383 The Brain and Space 0.3288 0.3164 0.3165 0.3023 0.3163 Network Analysis in Systems Biology 0.4053 0.3984 0.4034 0.3881 0.3914
automata 0.5575 0.5647 0.5621 0.556 0.5384
Natural Language Processing 2013 0.2466 0.2544 0.2472 0.2545 0.2644 Beginning Game Programming with C 0.233 0.2402 0.2538 0.238 0.2213 Climate Change 0.221 0.2185 0.2157 0.2204 0.2262 Journalism Skills for Engaged Citizens 0.4183 0.4351 0.4248 0.426 0.4136 Algorithms Design and Analysis Part 1 0.3293 0.3508 0.3495 0.3398 0.3105 Algorithms Design and Analysis Part 2 0.3405 0.3609 0.3573 0.3533 0.3465 Introduction to Chemistry Reactions and
Ratios
0.217 0.2103 0.2094 0.2126 0.2215 Genomic and Precision Medicine 0.2295 0.2368 0.2358 0.2349 0.2543 Epigenetic Control of Gene Expression 0.2832 0.3313 0.3268 0.3027 0.3039 Take the Lead on Healthcare Quality
Improvement
0.4142 0.4251 0.4243 0.3935 0.4163 Surviving Disruptive Technologies 0.4119 0.3582 0.3734 0.345 0.3907 Caries Management by Risk Assessment
CAMBRA
0.351 0.403 0.416 0.3474 0.3233
179
Computational Neuroscience 0.3232 0.3066 0.3058 0.306 0.3086 Introduction to Data Science 0.2965 0.3001 0.3074 0.2968 0.2996 Discrete Optimization 0.2971 0.3032 0.298 0.298 0.2914 Foundations of Virtual Instruction 0.1972 0.2559 0.2818 0.2278 0.2919 Virology I How Viruses Work 0.2312 0.2453 0.2408 0.2456 0.221 Edx Introduction to Computer Science 0.2507 0.2665 0.272 0.2588 0.2575 edx Big Data in Education 0.4801 0.4934 0.4986 0.4723 0.4827 edx Cellular mechanisms of brain
function
0.1801 0.175 0.1678 0.1771 0.1881 edx Introduction to Programming with
Java Part 1 Starting to Code with Java
0.1464 0.1406 0.1373 0.1356 0.1315 edx Programming Basics 0.3981 0.3887 0.3912 0.3878 0.3943 總平均 0.314 0.321 0.323 0.3101 0.3145
平均 F1 值:與一般 LDA 摘要方法呈現顯著差異
平均 F1 值:與 PLSA 摘要方法呈現顯著差異
紅色字體平均 F1 值:平均 F1 值未比一般 LDA 摘要方法或 PLSA 摘要方法高
在表 85 的結果中除了 Social Network Analysis 課程、Malicious Software and its Underground Economy Two Sides to Every Story 課程、The Brain and Space 課程、
Network Analysis in Systems Biology 課程、Natural Language Processing 2013 課程、
Climate Change 課程、Introduction to Chemistry Reactions and Ratios 課程、Genomic and Precision Medicine 課程、Surviving Disruptive Technologies 課程、Computational Neuroscience 課程、Foundations of Virtual Instruction 課程、edx Cellular mechanisms of brain function 課程、edx Introduction to Programming with Java Part 1 Starting to Code with Java 課程和 edx Programming Basics 課程的 TF 摘要方法、TF-ISF 摘要 方法和 TF-SF 摘要方法平均 F1 值皆比一般 LDA 摘要方法或 PLSA 摘要方法低,
其餘課程的 TF 摘要方法、TF-ISF 摘要方法和 TF-SF 摘要方法平均 F1 值皆至少 有一個摘要方法比一般 LDA 摘要方法和 PLSA 摘要方法高。
表 86. 與動詞 LDA 摘要方法和名詞 LDA 摘要方法比較結果
課程名稱 動詞
LDA
TF TF-IS F
TF-SF 名詞
LDA The Hardware/Software Interface 0.3785 0.3707 0.3763 0.3619 0.3693 Social Network Analysis 0.3238 0.3168 0.3168 0.3106 0.3221 Web Application Architectures 0.4833 0.4881 0.4885 0.4682 0.4938 Audio Signal Processing for Music
Applications
0.2191 0.225 0.2286 0.2096 0.2212
180 Malicious Software and its Underground Economy Two Sides to Every Story
0.4251 0.4167 0.4274 0.4129 0.4411 Big Data Science with the
BD2K-LINCS Data Coordination and Integration Center
0.2981 0.3163 0.314 0.2953 0.3084
Experimental Methods in Systems Biology
0.1081 0.0965 0.1016 0.0868 0.0865 Dynamical Modeling Methods for
Systems Biology
0.3823 0.3847 0.39 0.3675 0.3634 The Brain and Space 0.3019 0.3164 0.3165 0.3023 0.3337 Network Analysis in Systems Biology 0.3885 0.3984 0.4034 0.3881 0.3848
automata 0.5445 0.5647 0.5621 0.556 0.559
Natural Language Processing 2013 0.2449 0.2544 0.2472 0.2545 0.2521 Beginning Game Programming with C 0.2425 0.2402 0.2538 0.238 0.2711 Climate Change 0.2085 0.2185 0.2157 0.2204 0.2253 Journalism Skills for Engaged Citizens 0.4352 0.4351 0.4248 0.426 0.4249 Algorithms Design and Analysis Part 1 0.3123 0.3508 0.3495 0.3398 0.2956 Algorithms Design and Analysis Part 2 0.3583 0.3609 0.3573 0.3533 0.3475 Introduction to Chemistry Reactions and
Ratios
0.2447 0.2103 0.2094 0.2126 0.2274 Genomic and Precision Medicine 0.2307 0.2368 0.2358 0.2349 0.2451 Epigenetic Control of Gene Expression 0.2927 0.3313 0.3268 0.3027 0.2888 Take the Lead on Healthcare Quality
Improvement
0.4256 0.4251 0.4243 0.3935 0.4295 Surviving Disruptive Technologies 0.3743 0.3582 0.3734 0.345 0.3847 Caries Management by Risk Assessment
CAMBRA
0.3921 0.403 0.416 0.3474 0.3702 Computational Neuroscience 0.3069 0.3066 0.3058 0.306 0.2996 Introduction to Data Science 0.3067 0.3001 0.3074 0.2968 0.3073 Discrete Optimization 0.2824 0.3032 0.298 0.298 0.2928 Foundations of Virtual Instruction 0.1633 0.2559 0.2818 0.2278 0.2472 Virology I How Viruses Work 0.2322 0.2453 0.2408 0.2456 0.2286 Edx Introduction to Computer Science 0.2591 0.2665 0.272 0.2588 0.2652 edx Big Data in Education 0.4921 0.4934 0.4986 0.4723 0.4854 edx Cellular mechanisms of brain
function
0.1744 0.175 0.1678 0.1771 0.1794 edx Introduction to Programming with
Java Part 1 Starting to Code with Java
0.1276 0.1406 0.1373 0.1356 0.1495 edx Programming Basics 0.3932 0.3887 0.3912 0.3878 0.3927 總平均 0.3137 0.321 0.323 0.3101 0.318
平均 F1 值:與動詞 LDA 摘要方法呈現顯著差異
平均 F1 值:與名詞 LDA 摘要方法呈現顯著差異
紅色字體平均 F1 值:平均 F1 值未比動詞 LDA 摘要方法或名詞 LDA 摘要方法高
在表 86 的結果中除了 The Hardware/Software Interface 課程、Social Network
Analysis 課程、Web Application Architectures 課程、Malicious Software and its
Underground Economy Two Sides to Every Story 課程、Experimental Methods in
181
Systems Biology 課程、The Brain and Space 課程、Climate Change 課程、Journalism Skills for Engaged Citizens 課程、Introduction to Chemistry Reactions and Ratios 課 程、Genomic and Precision Medicine 課程、Take the Lead on Healthcare Quality Improvement 課 程 、 Surviving Disruptive Technologies 課 程 、 Computational Neuroscience 課程、Foundations of Virtual Instruction 課程、edx Cellular mechanisms of brain function 課程、edx Introduction to Programming with Java Part 1 Starting to Code with Java 課程和 edx Programming Basics 課程的 TF 摘要方法、TF-ISF 摘要 方法和 TF-SF 摘要方法平均 F1 值皆比動詞 LDA 摘要方法和名詞 LDA 摘要方法 低,其餘課程的 TF 摘要方法、TF-ISF 摘要方法和 TF-SF 摘要方法平均 F1 值至 少有一個摘要方法比動詞 LDA 摘要方法和名詞 LDA 摘要方法高。
表 87. 與線上摘要器 SweSum 比較結果
課程名稱 TF TF-ISF TF-SF SweSum
The Hardware/Software Interface 0.3707 0.3763 0.3619 0.3797 Social Network Analysis 0.3168 0.3168 0.3106 0.3371 Web Application Architectures 0.4881 0.4885 0.4682 0.5109 Audio Signal Processing for Music
Applications
0.225 0.2286 0.2096 0.2495 Malicious Software and its Underground
Economy Two Sides to Every Story
0.4167 0.4274 0.4129 0.439 Big Data Science with the BD2K-LINCS Data
Coordination and Integration Center
0.3163 0.314 0.2953 0.2951 Experimental Methods in Systems Biology 0.0965 0.1016 0.0868 0.091 Dynamical Modeling Methods for Systems
Biology
0.3847 0.39 0.3675 0.3794 The Brain and Space 0.3164 0.3165 0.3023 0.3422 Network Analysis in Systems Biology 0.3984 0.4034 0.3881 0.3993
automata 0.5647 0.5621 0.556 0.5394
Natural Language Processing 2013 0.2544 0.2472 0.2545 0.2429 Beginning Game Programming with C 0.2402 0.2538 0.238 0.2662
Climate Change 0.2185 0.2157 0.2204 0.2223
Journalism Skills for Engaged Citizens 0.4351 0.4248 0.426 0.4255 Algorithms Design and Analysis Part 1 0.3508 0.3495 0.3398 0.3238 Algorithms Design and Analysis Part 2 0.3609 0.3573 0.3533 0.3606 Introduction to Chemistry Reactions and
Ratios
0.2103 0.2094 0.2126 0.2034 Genomic and Precision Medicine 0.2368 0.2358 0.2349 0.2358 Epigenetic Control of Gene Expression 0.3313 0.3268 0.3027 0.3349 Take the Lead on Healthcare Quality 0.4251 0.4243 0.3935 0.3841
182 Improvement
Surviving Disruptive Technologies 0.3582 0.3734 0.345 0.3947 Caries Management by Risk Assessment
CAMBRA
0.403 0.416 0.3474 0.4158 Computational Neuroscience 0.3066 0.3058 0.306 0.3261 Introduction to Data Science 0.3001 0.3074 0.2968 0.3121 Discrete Optimization 0.3032 0.298 0.298 0.2944 Foundations of Virtual Instruction 0.2559 0.2818 0.2278 0.2364 Virology I How Viruses Work 0.2453 0.2408 0.2456 0.2431 Edx Introduction to Computer Science 0.2665 0.272 0.2588 0.249 edx Big Data in Education 0.4934 0.4986 0.4723 0.506 edx Cellular mechanisms of brain function 0.175 0.1678 0.1771 0.1721 edx Introduction to Programming with Java
Part 1 Starting to Code with Java
0.1406 0.1373 0.1356 0.1426 edx Programming Basics 0.3887 0.3912 0.3878 0.3733
總平均 0.321 0.323 0.3101 0.3221
平均 F1 值:與線上摘要器 SweSum 呈現顯著差異
紅色字體平均 F1 值:平均 F1 值未比線上摘要器 SweSum 高
在表 87 的結果中除了 The Hardware/Software Interface 課程、Social Network Analysis 課程、Web Application Architectures 課程、Malicious Software and its Underground Economy Two Sides to Every Story 課程、The Brain and Space 課程、
Beginning Game Programming with C 課程、Climate Change 課程、Epigenetic Control of Gene Expression 課程、Take the Lead on Healthcare Quality Improvement 課程、Surviving Disruptive Technologies 課程、Computational Neuroscience 課程、
Introduction to Data Science 課程、edx Big Data in Education 課程和 edx Introduction to Programming with Java Part 1 Starting to Code with Java 課程的 TF 摘要方法、
TF-ISF 摘要方法和 TF-SF 摘要方法平均 F1 值皆比線上摘要器 SweSum 低,其餘 課程的 TF 摘要方法、TF-ISF 摘要方法和 TF-SF 摘要方法平均 F1 值皆至少有一 個摘要方法比線上摘要器 SweSum 高。
未刪章節使用提示詞出現次數乘上 TF 權重加權提示詞時與一般 LDA 摘要
方法和 PLSA 摘要方法比較結果如表 88 所示,與動詞 LDA 摘要方法和名詞 LDA
摘要方法比較結果如表 89 所示,與線上摘要器 SweSum 比較結果如表 90 所示。
183
表 88. 與一般 LDA 摘要方法和 PLSA 摘要方法比較結果
課程名稱 一般
LDA
TF TF-ISF TF-SF PLSA
The Hardware/Software Interface
0.3747 0.3783 0.38 0.3764 0.3655 Social Network Analysis 0.3334 0.328 0.3275 0.3153 0.3072 Web Application Architectures 0.4834 0.4964 0.4935 0.4895 0.4785 Audio Signal Processing for
Music Applications
0.2146 0.2349 0.226 0.2379 0.2033 Malicious Software and its
Underground Economy Two Sides to Every Story
0.4268 0.4422 0.4385 0.4441 0.4275
Big Data Science with the
BD2K-LINCS Data
Coordination and Integration Center
0.2789 0.3264 0.3155 0.3128 0.3132
Experimental Methods in Systems Biology
0.0883 0.0986 0.1015 0.1009 0.094 Dynamical Modeling Methods
for Systems Biology
0.3757 0.3953 0.3899 0.3946 0.383 The Brain and Space 0.3288 0.3122 0.3139 0.3113 0.3163 Network Analysis in Systems
Biology
0.4053 0.4118 0.4104 0.3972 0.3914
automata 0.5575 0.5638 0.5607 0.5604 0.5384
Natural Language Processing 2013
0.2466 0.2562 0.246 0.2658 0.2644
Beginning Game
Programming with C
0.233 0.2799 0.2705 0.2625 0.2213 Climate Change 0.221 0.2136 0.2109 0.2151 0.2262 Journalism Skills for Engaged
Citizens
0.4183 0.4376 0.429 0.4454 0.4136 Algorithms Design and
Analysis Part 1
0.3293 0.3505 0.3441 0.3565 0.3105 Algorithms Design and
Analysis Part 2
0.3405 0.3676 0.3615 0.3633 0.3465 Introduction to Chemistry
Reactions and Ratios
0.217 0.2097 0.2195 0.2127 0.2215 Genomic and Precision
Medicine Surviving Disruptive
Technologies
0.4119 0.3924 0.3954 0.3753 0.3907 Caries Management by Risk
Assessment CAMBRA
0.351 0.4347 0.4345 0.4081 0.3233 Computational Neuroscience 0.3232 0.3129 0.3125 0.3047 0.3086 Introduction to Data Science 0.2965 0.3045 0.2998 0.3052 0.2996 Discrete Optimization 0.2971 0.3056 0.3019 0.3038 0.2914 Foundations of Virtual
Instruction
0.1972 0.2814 0.2807 0.2871 0.2919