際際滷

際際滷Share a Scribd company logo

Real conversations with artificial intelligence

B
Byungjoon Kim

豈覺螻殊 襯 語危朱 覿 覦.

1 of 12
Real conversations with artificial intelligence: A comparison between human-human and human-chatbot + Computers in Human Behavior - Jennifer Hill, W.Randolph Ford, and Ingrid G.rarreras /蟾覲譴 X 2017 Spring
朱語 覲 伎 覈 覿 磯 蟆 2螳讌襦 讌 / 覈 : 覈螳 ex) 譯朱, 誤磯薫 譟一, 覯 谿剰鍵 觜覈 : 覈螳 ex) ° 郁規 觜る 覈 り 豐 襷豢讌襷 れ 覺螻殊 Interaction り , ろ °危, 觜覈 襯 Control蟆 譴讌 蟾?
Intro Intelligent agent 蠍一 螻 覿蟲螻 襷 殊企螻 . Intelligent agent 讌 蠍一 牛 企伎 CMC(Computer- mediated communication)螳 譴 郁規 覿手 螻 . 郁規れ (Face to face, Instant messaging(IM), text messages) 郁規螳 . 覩碁ゼ 螻 焔レ レる 蠍一 郁規れ . Turing Test 蠍磯 貉危一 郁規 讌襷 覈 螳 ｸル 讌. But, 豈覺 襷 覿蟲螻 豈覺螻殊 郁規 .
Human-human online conversations Human-chatbot conversations 企至 殊讌 覲伎 Purpose * Chatbot : Machine conversation systems interact with human users via natural conversational language
Method 一危 讌 Human-human conversations : , 朱覿 伎襷 殊企 IM 一危 讌 205覈 谿瑚襦覿 894螳 IM Human-chatbot conversations : Cleverbot Rollo Carpenter襦覿 Cleverbot 一危磯ゼ 覦 2140螳 Cleverbot 蠍磯
Method 一危 讌 一危 豌襴 10 螳 覩碁 覃讌 蟇 企 讌 襷 螳 一危一 ろ蟆 100螳 襯
Method 一危 讌 一危 豌襴 一危 覿 7螳 覲襯 覿 : Words per conversation, Messages per conversation, Average number of words per message, Type/token ratio, IM shorthand, emotions, occurrences of profanity 14螳 LIWC(Linguistic Inquiry and. Word Count) 覲 覿 : Word count, Words per sentence, Percentages of words greater than six letters, LIWC dictionary words, Function words, Pronouns, Articles, Swear words, Social words, Affective words, Positive emotion words, Negative emotion words, Sexual words, Words of assent
Linguistic variables analyzed 1. Words White space襦 蟲覿, 企一企 蟲 覦一 2. Type-token ratio 豌 word 譴 unique word 觜 3. Profanity Banbuilder.com 焔 襴ろ ex) 觜, 語谿覲, 煙 4. Shorthand Internetslang.com 焔 襴ろ ex) BOGO, SLAP 5. Emotions Internetslang.com 焔 襴ろ ex) ^_^ , :) , >//< 6. Messages 蠍語企 伎 蟯 ° 一危 蠍一, 覦覲 ° 一危磯 覃讌襦
Result Linguistic variables 7螳 覲 譴 4螳 覲(Messages per conversation, Words per message, Type/token ratio, Frequency of profanity)螳 覩誤 谿企ゼ 覲伎. 螳 觜 豈覺螻殊 2覦磯慨 襷 覃讌襯 覲企碁 覃讌 蠍語企 螳 螳 2覦 蠍碁. 螳 襷 企れ . 螳 覲企 豈覺螻殊 襷 觜企ゼ .
Result LIWC variables LIWC 覲 譴 8螳 覲(WPS, Words more than Six-letter, Pronoun, Swear, Social, Pos emot, Neg emot, Sexual)螳 覩誤 谿企ゼ 覲伎. 螳 豈覺螻殊 覲企 覓語 伎 螳 10覦 螳蟾 襷. 螳 豌螳 襷 企ゼ . 豈覺 螳 覈襯 襷 . 豈覺 螳 觜企ゼ . 豈覺 螳 企ゼ . 螳 蠍 企一 覦覃, 豈覺 螳 覿 企一 . 豈覺 螳 煙 企ゼ .
Future Research 1. 覓碁 蟲譟 觜蟲, 螳 覓語レ 螳 覃讌 覓語レ 螳 覃讌 2. 覃讌襯 覲企企 螳 覦 豺覓旧 蠍語 3. 譯殊螳 覦
Discussion 譯殊 覯襯 譬 蠏狩覃 企蟾? 譯殊 磯ジ 覲螳 蠍瑚? 豈覺螻殊 螻殊 碁 螳 覦 覲 讌 蟾?

More Related Content

Real conversations with artificial intelligence

  • 1. Real conversations with artificial intelligence: A comparison between human-human and human-chatbot + Computers in Human Behavior - Jennifer Hill, W.Randolph Ford, and Ingrid G.rarreras /蟾覲譴 X 2017 Spring
  • 2. 朱語 覲 伎 覈 覿 磯 蟆 2螳讌襦 讌 / 覈 : 覈螳 ex) 譯朱, 誤磯薫 譟一, 覯 谿剰鍵 觜覈 : 覈螳 ex) ° 郁規 觜る 覈 り 豐 襷豢讌襷 れ 覺螻殊 Interaction り , ろ °危, 觜覈 襯 Control蟆 譴讌 蟾?
  • 3. Intro Intelligent agent 蠍一 螻 覿蟲螻 襷 殊企螻 . Intelligent agent 讌 蠍一 牛 企伎 CMC(Computer- mediated communication)螳 譴 郁規 覿手 螻 . 郁規れ (Face to face, Instant messaging(IM), text messages) 郁規螳 . 覩碁ゼ 螻 焔レ レる 蠍一 郁規れ . Turing Test 蠍磯 貉危一 郁規 讌襷 覈 螳 ｸル 讌. But, 豈覺 襷 覿蟲螻 豈覺螻殊 郁規 .
  • 4. Human-human online conversations Human-chatbot conversations 企至 殊讌 覲伎 Purpose * Chatbot : Machine conversation systems interact with human users via natural conversational language
  • 5. Method 一危 讌 Human-human conversations : , 朱覿 伎襷 殊企 IM 一危 讌 205覈 谿瑚襦覿 894螳 IM Human-chatbot conversations : Cleverbot Rollo Carpenter襦覿 Cleverbot 一危磯ゼ 覦 2140螳 Cleverbot 蠍磯
  • 6. Method 一危 讌 一危 豌襴 10 螳 覩碁 覃讌 蟇 企 讌 襷 螳 一危一 ろ蟆 100螳 襯
  • 7. Method 一危 讌 一危 豌襴 一危 覿 7螳 覲襯 覿 : Words per conversation, Messages per conversation, Average number of words per message, Type/token ratio, IM shorthand, emotions, occurrences of profanity 14螳 LIWC(Linguistic Inquiry and. Word Count) 覲 覿 : Word count, Words per sentence, Percentages of words greater than six letters, LIWC dictionary words, Function words, Pronouns, Articles, Swear words, Social words, Affective words, Positive emotion words, Negative emotion words, Sexual words, Words of assent
  • 8. Linguistic variables analyzed 1. Words White space襦 蟲覿, 企一企 蟲 覦一 2. Type-token ratio 豌 word 譴 unique word 觜 3. Profanity Banbuilder.com 焔 襴ろ ex) 觜, 語谿覲, 煙 4. Shorthand Internetslang.com 焔 襴ろ ex) BOGO, SLAP 5. Emotions Internetslang.com 焔 襴ろ ex) ^_^ , :) , >//< 6. Messages 蠍語企 伎 蟯 ° 一危 蠍一, 覦覲 ° 一危磯 覃讌襦
  • 9. Result Linguistic variables 7螳 覲 譴 4螳 覲(Messages per conversation, Words per message, Type/token ratio, Frequency of profanity)螳 覩誤 谿企ゼ 覲伎. 螳 觜 豈覺螻殊 2覦磯慨 襷 覃讌襯 覲企碁 覃讌 蠍語企 螳 螳 2覦 蠍碁. 螳 襷 企れ . 螳 覲企 豈覺螻殊 襷 觜企ゼ .
  • 10. Result LIWC variables LIWC 覲 譴 8螳 覲(WPS, Words more than Six-letter, Pronoun, Swear, Social, Pos emot, Neg emot, Sexual)螳 覩誤 谿企ゼ 覲伎. 螳 豈覺螻殊 覲企 覓語 伎 螳 10覦 螳蟾 襷. 螳 豌螳 襷 企ゼ . 豈覺 螳 覈襯 襷 . 豈覺 螳 觜企ゼ . 豈覺 螳 企ゼ . 螳 蠍 企一 覦覃, 豈覺 螳 覿 企一 . 豈覺 螳 煙 企ゼ .
  • 11. Future Research 1. 覓碁 蟲譟 觜蟲, 螳 覓語レ 螳 覃讌 覓語レ 螳 覃讌 2. 覃讌襯 覲企企 螳 覦 豺覓旧 蠍語 3. 譯殊螳 覦
  • 12. Discussion 譯殊 覯襯 譬 蠏狩覃 企蟾? 譯殊 磯ジ 覲螳 蠍瑚? 豈覺螻殊 螻殊 碁 螳 覦 覲 讌 蟾?