關鍵字廣告是指廣告贊助商付費於網路搜尋引擎後，搜尋引擎根據使用者的關鍵字將贊助商的廣告呈現在搜尋結果的旁邊，而關鍵字廣告也是目前網路搜尋引擎最大的利潤來源之一。對於關鍵字廣告來說，其目的就是要讓使用者能看見，並且點擊此廣告。因此，搜尋引擎如何推薦一個合適的廣告可以讓使用者覺得該則廣告與他輸入的關鍵字是有相關的仍是一個很重要的議題。然而大多數的搜尋引擎在使用者輸入關鍵字後，依舊提供大量不相關的廣告給予使用者，造成使用者的困擾。其原因是因為搜尋引擎依據廣告商所買的廣告詞彙來做匹配，並且沒有考慮到使用者真正的需求下給予廣告。因此，本論文提出任務導向廣告模型，希望藉此模型可以改善關鍵字廣告推薦的缺點。
我們的任務導向廣告模型包含了三個子模型，分別為問句任務辨識、搜尋意圖&情緒擷取與廣告推薦。首先我們先要辨識使用者他背後的複雜搜尋任務是什麼；接者找出該任務下包含哪些的搜尋意圖與使用者可能的情緒，最後再結合廣告熱門度與廣告意見配對找出適合的廣告與其他可能相關的廣告推薦給使用者。
由實驗結果證實我們提出的模型與“奇摩關鍵字廣告”比較可以得到較高的效能，並且能推薦較相關且滿足使用者其他需求的廣告。另外我們也證實我們找出的使用者情緒，廣告熱門度與廣告意見配對也能改善關鍵字廣告的缺點及排名。

The phenomenon of Keyword Advertising – where advertisers pay a fee to Internet search engines to be displayed alongside organic web search results – is gaining ground as the largest source of revenues for search engines. Since users are more likely to click ads that are relevant to their query, it is crucial for search engine to deliver the right ads for the query and the order in which they are displayed. However, search engines are still provide many ads that are irrelevant to users for keyword-based and just consider a query only corresponding to a simple user task. In this paper, we proposed a Task-based Advertisement Model to solve these problems and aim to provide suitable ads for users.
We introduce three sub-models which contained in our Task-based Advertisement Model. First, Query Task Identification which finds out users’ complex search task based on users’ query. Second, Search Goal & Emotion Extraction is used to find other users’ needs and predict users’ emotion to obtain more appropriate ads. Finally, Ad Recommend adds extra two ad features, opinion pair and popularity to generate final ad list to users.
The experiment results verify the proposed model that can produce higher relevant ads than “Yahoo” keyword ads as well as deliver other related ads for the complex search task. We also demonstrate that user emotion, opinion pair and popularity can improve the shortage and ranking of advertising.

[1] Agichtein, E., Brill, E., Dumais, S., and Ragno, R. Learning user interaction models for predicting web search result preferences. In Proc. of SIGIR, 2006.
[2] Beeferman, D. and Berger, A. Agglomerative clustering of a search engine query log. In Proc. of KDD ‘00, 407-416, 2000.
[3] Broder, A. A taxonomy of web search. In Proc. of SIGIR Forum, 2002.
[4] Brenes, D. J., Gayo-Avello, D., and Pérez-González, K. Survey and evaluation of query intent detection methods. In Proc. of WSCD workshop, pages 1-7, 2009.
[5] Cui, J., Liu, H., Yan, J., Ji L., Jin R., He, J., Gu, Y., Chen, Z., and Du, X. Multi-view Random Walk Framework for Search Task Discovery from Click-through Log. In Proc. of CIKM’11.
[6] Dice, Lee R. Measures of the Amount of Ecologic Association Between Species. 1945.
[7] Elkan C. Log-linear models and conditional random fields. In Proc. of CIKM, 2008
[8] Ghose, A., and Yang, S., An empirical analysis of sponsored search performance in search engine advertising. In Proc. Conference’08.
[9] Guo, Q. and Agichtein, E. Ready to buy or just browsing? detecting web searcher goals from interaction data. In Proc. of SIGIR, 2010.
[10] Hubbard, L. R. The Emotional Tone Scale, New Era Publications International, 2001.
[11] Kotov, A., Bennett, P. N., White, R. W., Dumais, S. T., and Teevan, J. Modeling and analysis of cross-session search tasks. In Proc. of SIGIR, 2011.
[12] Lee, U., Liu, Z., and Cho, J. Automatic identification of user goals in web search. In Proc. of WWW, 2005.
[13] Liao, Y., and Chang, S. Advertisement before the show: The Effect of Ad-evoked and consequent media-evoked emotion to the ads’ impression. Chinese Communication Society, Taiwan, 2011.
[14] Lin, T., Pantel, P., Gamon, M., Kannan, A., Fuxman, A. Active Objects: Actions for Entity-Centric Search. In Proc, WWW’12.
[15] Lin, S., and Lin, C. Emotions touch my heart：The Influence of Advertising Emotion and Product Involvement on Advertising Effectiveness from Approach-Avoidance Theory. Chinese Communication Society, Taiwan, 2011.
[16] Mishne, G., and Glance, N. Predicting Movie Sales from Blogger Sentiment. In Proc. of AAAI, 2006.
[17] Ravi, S., Broder, A., Gabrilovich, E., Josifovski, V., Pandey, S., and Pang, B. Automatic generation of bid phrases for online advertising. In Proc. of WSDM’10.
[18] Sørensen, T. A method of establishing groups of equal amplitude in plant sociology based on similarity of species and its application to analyses of the vegetation on Danish commons. 1948.
[19] Salton, G., and McGill, M. J. Introduction to modern information retrieval. 3rd ed. New York: McGraw-Hill. 1983.
[20] Singhal, A. Modern Information Retrieval: A Brief Overview. In Prof, IEEE, 2001.
[21] Wang, w., Huang, S., and Liu, Y. Emotional Valence and Arousal Analysis of Consumer Goods Advertising Design. Journal of Design,Taiwan, page 45-67, 2012.
[22] Watson, D., Tellegen, A. Toward a consensual structure of mood. Psychological Bulletin. 219-235,1985.
[23] Zhu, Y., Wang, G., Yang, J., Wang, D., Yan, J., Hu, J., and Chen, Z. Optimizing search engine revenue in sponsored search. In Proc. of SIGIR, 2009.
[24] 吳松柏, 藉由Yahoo知識+的隱含需求提出事件驅動情緒性需求的部落格廣告方式, 碩士論文, 國立成功大學資訊工程所, 2010
[25] 許榮展, 利用具有情緒意圖的商品多面向評價以分析部落客購物需求,碩士論文, 國立成功大學資訊工程所, 2010
[26] 廖顯忠, 分析隱含於部落格文章的事件、情緒與需求, 碩士論文, 國立成功大學資訊工程所, 2010