任何由所述需求驅動的，並希望客戶組織已經明白從任何媒體客戶行為。 Twitter正在成為世界上最流行的社交媒體之一。如今，微博的很多第三方應用已經讓客戶自由地隨時隨地分享他們的意見。然而，這是很難理解客戶行為的變化，因為鳴叫被張貼在不穩定的可能性。因此，可視化客戶的情感行為的時間模式的Twitter可以發揮在決策至關重要的作用。現有的數據可視化工具，如D3.js，激勵我們去開發，並探討在二維的可視化的Twitter數據的時間維度。

我們提出的系統由三個主要部分組成：文本分析，可視化展示和交互。該輸入數據是從收集的微實時和直接存儲在我們的鳴叫數據庫。我們分離採集數據為兩個步驟。首先，我們檢索到一些特定國家的所有微博使用的地理邊界框，並存儲在我們的鳴叫的數據庫，那麼，我們建立了一個關鍵字過濾，“iPhone”，以獲取已準備好要處理的數據集。在此研究中使用Twitter的數據是從整個美國，日本，印尼和台灣收集，因為知道不同的位置可能會導致不同客戶的行為。之前呈現的可視化，以確定屬性，哪些詞被認為是正面或負面意見進行數據預處理。樸素貝葉斯分類器的文本然後用一組包含看來特徵所選詞語的開發。

最後，我們所提出的系統中使用的情緒分析的結果也可以在天/小時和週/天熱圖，交互圖形流和上下文集中通過塗刷可視化可視化。實驗結果表明，該系統可以探索和分析客戶的情感行為時空格局。

Any organization that is driven by the needs and wants of customers has to understand customer behavior from any media. Twitter is becoming one of the most popular social media in the world. Nowadays, many third-party applications of Twitter have been making customers freely to share their opinions anywhere and anytime. However, it is difficult to understand the change of customer behavior, because the tweet is posted in erratic likelihood. Therefore, visualizing the time pattern of customer emotion behavior on Twitter can play a crucial role in decision-making. Available data visualizing tools, such as D3.js, motivate us to develop and to explore time dimension of Twitter data in 2D visualization.

Our proposed system is composed of three main parts: text analysis, visualization presentation, and interaction. The input data was collected from Twitter in real-time and stored directly in our tweet database. We separated acquiring data into two steps. First, we retrieved all tweets from several certain countries by using geographical bounding-box and stored in our tweet database, then we set up a keyword filter, “iPhone”, to get the dataset that is ready to be processed. Twitter data used in this research was collected from the entire U.S.A., Japan, Indonesia, and Taiwan, since knowing different location may result different customer behavior. Prior to present visualization, data preprocessing was performed in order to determine attributes, which words are considered as positive or negative opinion. Naïve Bayes text classifier was then developed using a set of the chosen words containing opinion features.

Finally, our proposed system used the result of sentiment analysis to be visualized in day / hour and week / day heat map, interactive stream graph, and context focus via brushing visualization. The results show that our proposed system can explore and analyze the temporal pattern of customer emotions behavior.

[1] A. Pak, a. P. (2010). Twitter as a Corpus for Sentiment Analysis and Opinion Mining. Proc. Of LREC.
[2] Alper, B., Yang, H., Haber, E., & & Kandogan, E. (n.d.). OpinionBlocks: VIsualizing Consumer Reviews.
[3] Ben, W., & Karaa, A. (July 2013). A New Stemmer to Improve Information Retrieval. International Journal of Network Security & Its Applications (IJNSA), 4.
[4] Brown, J. S., & and Duguid, P. (2002). The Social Life of Information. Boston, USA: Hardvard Business School Press.
[5] Chernov, J. (2009). Social Media Playbook: Everything Your Company Needs to Know to Succeed on the Social Web. In J. Chernov, Social Media Playbook: Everything Your Company Needs to Know to Succeed on the Social Web (pp. 16-19). Eloqua Corporation.
[6] D., C., Manning, Rahavan, P., & and Schutze, H. (2008). Introduction to Information Retrieval. Cambridge University Press.
[7] Dommati, S. J., Agrawal, R., Reddy, R. M., & and Kamath, S. (2012). Comparison of Event Model using Naive Bayes Approach. International Conference of Recent Trends in Computer and Information Engineering (ICRTCIE), 8-12.
[8] Duan, D., Qian, W., Shi, L., Pan, S., & and Lin, C. (2012). VISA: A VIsual Sentiment Analysis System. VINCI, Hangzhou, China.
[9] Go, A., Bhayani, R., & Huang, L. (2009). Twitter Sentiment Classification using Distant Supervision. CS224N Stanford, 1-12.
[10] Hao, M. C., Rohrdantz, C., Janetzko, H., Keim, D. A., Dayal, U., Haug, L. e., . . . and Stoffel, F. (2013). Visual Sentiment Analysis of Customer Feedback Streams Using Geo-Temporal Term Associations. Information Visualization 12 SAGE, 273-290.
40
[11] Hogenboom, A., Bal, D., & Frasincar, F. (2013). Exploiting Emoticons in Sentiment Analysis. SAC ACM.
[12] Huang, J., Cen, Z., & and Zheng, Q. (2010). Research and Implement of Chinese Text Classifier Based on Naive Bayes Method. IEEE 6th International Conference on Semantics, Knowledge, and Grids, 426-428.
[13] Itoh, M., Toyoda, M., & and Kitsuregawa, M. (2013). Visualizing Time-Varying Topics via Images and Texts for Inter-Media Analysis. 17th International Conference on Information Visualization , 568-576.
[14] Iwanaga, M., Tabata, T., & and Sakurai, K. (n.d.). Some Fitting of Naive Bayesian Spam Filtering for Japanase Environment.
[15] Kwak, H., Lee, C., Park, H., & and Moon, S. (2010). What is Twitter, a Social Network or a News Media. ACM International World Web Wide Conference Committee (IW3C2).
[16] Lipizzi, C., Iandoli, L., & Marquez, J. E. (2015). Extracting and Evaluating Conversational Patterns in Social Media: A Socio-Semantic Analysis of Customers' Reactions to the Launch of New Products Using Twitter Streams. International Journal of Information Management 35, Elsevier Ltd., 490-503.
[17] M. van Meeteren, A. P. (2009). Mapping Communities in Large Virtual Social Networks. Proocengs of the 1st International Forum on the Application and Management of Personal Electronic Information. Cambridge.
[18] Malik, S., Smith, A., Hawes, T., Papadatos, P., Li, J., Dunne, C., & and Shneiderman, B. (2013). TopicFlow: Visualizing Topic Alignment of Twitter Data Over Time. IEEE/ACM International Conference on Advances in Soical Networks Analysis and Mining, 720-726.
[19] Mira, J., & and Alvarez, J. R. (June 15-18, 2015). Mechanisms, Symbols, and Models Underlying Cognition. First International Work Conference on the Interplay Between Natural and Artificial Computation, IWINAC, Proocedings (p. 282). Las Palmas, Spain: Springer.
41
[20] Qi, Y., Shi, G., Yu, X., & Li, Y. (2015). Visualization in Media Big Data Analysis. ICIS IEEE.
[21] Semiocast. (2012). Retrieved from Semiocast Top 20 Countries: https://semiocast.com/en/publications/2012_01_31_Brazil_becomes_2nd_country_on_Twitter_superseds_Japan
[22] Tang, J., Liu, Z., & and Sun, M. (3013). THe Review of Text Visualization. Journal of Computer-Aided Desing & COmputer Graphics, 273-285.
[23] Tseng, C., Patel, N., Paranjape, H., Lin, T. Y., & and Teoh, S. (2012). Classifying Twitter Data with Naive Bayes Classifier. IEEE International Conference on Granular Computing.
[24] Tsytsarau, M., & and Palpanas, T. (2012). Survey on Mining Subjective Data on the Web. Data Mining and Knowledge Discovery.
[25] Twitter. (2006, July 6). Retrieved from Twitter Libraries: https://dev.twitter.com/docs/twitter-libraries
[26] V. Pandey, a. C. (n.d.). Sentiment Analysis of Micro blogs.
[27] Wang, A. H. (26-27 July 2010). Don't Follow Me: Spam Detection in Twitter. Proceedings of the 2010 International Conference on Security and Cryptography (SECRYPT), Piraeus.
[28] Wu, Y., Wei, F., Liu, S., Au, N., Cui, W., Zhou, H., & and Qu, H. (2010). OpinionSeer: Interactive Visualization of Hotel Customer Feedback. Transactions on Visualization and Computer Graphics, 1109-1118.
[29] Yamamoto, Y. (2013, September 9). http://twitter4j.org/ko/. Retrieved from http://twitter4j.org/ko/
[30] Yu, X.-R., Xiang, Z.-L., & Kang, D.-K. (2015). Classification of Chinese-To-English Translated Social Network Timeline using Naive Bayes. ICACT , 296-299.