Haiku generation, a subset of poetry generation, requires creativity, imagination and commonsense to produce high quality products. Humans possess the inherent ability to both produce and appreciate this, and any other form of art. They can instantly differentiate what is human made from what is not. The focus of this study is not to deal with the problem if computers can appreciate poetry (haikus), but rather if computers can produce poetry that can persuade humans to think they were made by another human being. To tackle this problem we use ConceptNet [1] [2], a multilingual knowledge base, which is conformed of nodes that contain words and phrases curated by people. These nodes have a common-sense relationship between them. We use this relationship to construct our haikus. Then, we developed a customized differential evolution genetic algorithm to improve the quality of these haikus. Finally, we ran a Turing test among a group of people to evaluate the quality of our haikus. Results from this test indicate machine-made haikus are of lesser quality compared to their human counterparts. However, an important proportion of machine generated haikus tended to confuse humans; and even obtained higher rating and surprise scores than their counterparts.

俳句一代，一代诗的一个子集，需要创造力，想象力和常识，以生产高品质的产品。人类拥有的固有能力都产生和欣赏这一点，和任何其他形式的艺术。它们可以立即区分什么是人类从什么是不可缺少的。这项研究的重点不是处理问题，如果计算机可以欣赏诗（俳句），而是计算机是否能够产生诗歌能说服人认为他们被另一个人发了言。为了解决这个问题，我们使用ConceptNet，多语言的知识基础，这是符合包含单词和策划人的短语节点。这些节点之间有一个常识性的关系。我们利用这种关系来构建我们的俳句。然后，我们开发了一个定制的差分进化遗传算法，以改善这些俳句的质量。最后，我们跑了图灵测试的一组人来评估我们的俳句的质量。结果从这个测试表明机器制造的俳句是质量较差相比，其人类对应。然而，机器产生的俳句的重要比例趋于迷惑人;甚至获得了较高的评价和惊喜的分数比他们的同行。

1. Introduction 1
2. Background 3
2.1 Haiku 3
2.2 ConceptNet 4
2.3 Differential Evolution Algorithms 10
2.4 NLTK (Natural Language Toolkit) 14
3. Related Work 17
3.1 Automatic Haiku Generation Using VSM 17
3.2 Gaiku: Generating Haiku with Word Associations Nouns 19
4. Model 21
4.1 Generate random haikus 22
4.2 Dimensions 24
4.3 Crossover 28
4.4 Mutation 32
5. Implementation 35
5.1 Random haiku generation process (Random Population) 35
5.2 Random haiku generation process (Random Population) 37
5.3 Differential Evolution 38
5.4 Mutation 39
5.5 Haiku Implementation 40
6. Analysis 41
7. Conclusions 47
8. Future Work 48
9. Bibliography 49

[1] R. S. a. C. Havasi, Representing General Relational Knowledge in ConceptNet 5, N. C. (. C. a. K. C. a. T. D. a. M. U. D. a. B. M. a. J. M. a. A. M. a. J. O. a. S. Piperidis, Ed., Istanbul, Turkey: European Language Resources Association (ELRA), 2012.
[2] P. S. H. Liu, "ConceptNet — A Practical Commonsense Reasoning Tool-Kit," BT Technology Journal, vol. 22, no. 4, pp. 211-226, 2004.
[3] B. Gascoigne, "HistoryWorld," 2001, ongoing. [Online]. Available: http://www.historyworld.net.
[4] H. O. a. M. M. N. Tosa, "Hitch haiku: An interactive supporting system for composing haiku poem.," Proc. of the 7th International Conference on Entertainment Computing, 2008.
[5] M. T. W. a. A. H. W. Chun, "Automatic Haiku Generation Using VSM," in 7th WSEAS Int. Conf. on APPLIED COMPUTER & APPLIED COMPUTATIONAL SCIENCE (ACACOS '08), Hangzhou, China, 2008.
[6] D. G. Y. G. a. M. E. Yael Netzer, "Gaiku : Generating Haiku withWord Associations Norms," in Proceedings of the NAACL HLT Workshop on Computational Approaches to Linguistic Creativity,, Boulder, Colorado, 2009.
[7] D. B. a. D. Merski, "Haiku poetry and metaphorical thought: An invention to interdisciplinary study," Creativity Research Journal,, no. 11, 1998.
[8] N. Project, "Natural Language Toolkit — NLTK 3.0 documentation," [Online]. Available: http://www.nltk.org/.
[9] R. Storn and K. Price, ""Differential evolution - a simple and efficient heuristic for global optimization over continuous spaces"," Journal of Global Optimization, pp. 341-359, 1997.
[10] I. Haiku Society of America, "Official Definitions of Haiku and Related Terms," Haiku Society of America, Inc., 18 September 2004. [Online]. Available: http://www.hsa-haiku.org/archives/HSA_Definitions_2004.html.
[11] C. J. Database, "View texts by #1 Matsuo Bashô (松尾芭蕉).," [Online]. Available: http://carlsensei.com/classical/index.php/author/view/1.
[12] C. H. Rob Speer, "ConceptNet 5," MIT Media Lab, [Online]. Available: http://conceptnet5.media.mit.edu/.
[13] G. G. a. W. Kahan, "Calculating the Singular Values and Pseudo-Inverse of a Matrix," Journal of the Society for Industrial and Applied Mathematics: Series B, Numerical Analysis, vol. 2, no. 2, pp. 205-224, 1965.
[14] L. Insight, "assoc-space," [Online]. Available: https://github.com/LuminosoInsight/assoc-space.
[15] http://www1.icsi.berkeley.edu/~storn/code.html, "Differential Evolution Homepage," University of California, Berkeley, [Online]. Available: view-source:http://www1.icsi.berkeley.edu/~storn/code.html.
[16] B. K. A. A. D.G. Mayer, "Differential evolution – an easy and efficient evolutionary algorithm for model," Agricultural Systems, vol. 83, no. 3, pp. 315-318, 2005.
[17] D. Yuret, "The problem of local maxima," 1 April 1997. [Online]. Available: view-source:http://www2.denizyuret.com/pub/aitr1569/node6.html.
[18] S. E. L. E. K. Bird, Natural Language Processing with Python., O’Reilly Media Inc., 2009.
[19] D. Zaharie, "A Comparative Analysis of Crossover Variants," in Proceedings of the International Multiconference on Computer Science and Information Technology, 2007.
[20] R. Speer, "Edges," GitHub, Inc, 27 March 2014. [Online]. Available: https://github.com/commonsense/conceptnet5/wiki/Edges.
[21] G. A. Miller, "WordNet: a lexical database for English," Communications of the ACM, vol. 38, no. 11, pp. 39-41, 1995.
[22] C. Fellbaum, WordNet: An Electronic Lexical Database, Cambridge: MIT Press, 1998.
[23] N. M. Y. Ahmad Razlan Yusoff, "A Refined Differential Evolution Algorithm for Improving the Performance of Optimization Process," in Informatics Engineering and Information Science, Springer Berlin Heidelberg, 2011, pp. 184 - 194.