社交平台上發表文章是年輕人最喜歡的活動。隨著電影產業的潛力，開發自動電影推薦引擎成為一個熱門話題。在社交媒體上，在共享相關預告片與關於日常生活在線社交平台的用戶生成文章的場景中，用戶傾向於選擇考慮其抒情主題的預告片。

為了解決上述問題，我們提出了一種基於關係 - 場景的預告片推薦系統，該系統可以通過分析抒情主題來推薦預告片列表到輸入文章。我們認為抒情主題是關係和情景的結合，是情節總結的主觀和客觀視角。通過利用關係情景數據庫（Extend-HowNet 作為知識庫），我們提取情節摘要和文章的關係和情景特徵。關係特徵表示為人物，情感，事件，地點和時間實體的實現。場景特徵表示為情感和事件實體的實現。

因此，我們使用關係和場景特徵提供更好的推薦結果，而不僅僅考慮其中一個特徵，最後我們的推薦系統在用戶偏好和系統性能的兩個實驗中都優於新的技術(W2V)。並且考慮了系統推出不同關係情境之預告片做評估。

The post articles on the social platform is the favorite activity of young people. With the potential of digital movie industry, developing automatic movie recommendation engines becomes a popular issue. On social media, in the
scenario of sharing related trailers with user-generated articles about daily life online social platforms, users tend to choose trailers considering their lyrical theme.

To solve the above problem, we present a Relationship-Scenario-based Trailer Recommendation System which can recommend list of trailers to an input article by analyzing lyrical theme. We consider lyrical theme as a combination of Relationship and Scenario, the subjective and objective perspective of plot summaries. By utilizing relationship-scenario Database (Extended-HowNet as Knowledge base), we extract relationship and scenario features of plot summaries and articles. Relationship feature is represented as character, emotion, event, location and time entity relation. And scenario feature is represented as emotion and event entity relation.

Consequently, we show that using both relationship and scenario features provide better recommendation results than merely consider one of the features, In the end our recommender system outperforms a novel W2V baseline in both experiments of user preference and system performance. Also we consider user preference on our system about different relationship class.

[1] Hadi Pouransari, Saman Ghili , “Deep learning for sentiment analysis of movie reviews ”, 2015
[2] Mihaela Sorostinean, Katia Sana, Sentiment Analysis on Movie Reviews , March 1, 2017
[3] Gaurav Arora1, Ashish Kumar2, MOVIE RECOMMENDATION SYSTEM BASED ON USERS’ SIMILARITY, IJCSMC, Vol. 3, Issue. 4, April 2014, pg.765 – 770
[4] Xuan-Son Vu† Seong-Bae Park†, Mining User/Movie Preferred Features Based on Reviews for Video Recommendation System, 2017
[5] Jennifer Golbeck1, Generating Predictive Movie Recommendations from Trust in Social Networks, 2006
[6] Tomas Mikolov, Kai Chen, Greg Corrado, and Jeffrey Dean. Efficient estimation of word representations in vector space. ICLR Workshop, 2013.
[7] Kim, Yoon. Convolutional neural net- works for sentence classification. arXiv preprint arXiv:1408.5882, 2014.
[8] Extended-HowNet, http://ehownet.iis.sinica.edu.tw/index.php
[9] Chen, K. J., Huang, S. L., Shih, Y. Y., & Chen, Y. J. 2005a. Extended-HowNet- A Representational Framework for Concepts. In Processing of OntoLex 2005 - Ontologies and Lexical Resources IJCNLP-05 Workshop, Jeju Island, South Korea.
[10] Huang, S. L., Chung, Y. S., & Chen, K. J. E-HowNet- an Expansion of HowNet. In Proceedings of 1st National HowNet Workshop, Beijing, China. 2008
[11] CKIP Group, Lexical Semantic Representation and Semantic Composition- An Introduction to E- How Net, CKIP Technical Report, 2013.
[12] Dong Zendong, and Qiang Dong, HowNet and the Computation of Meaning. World Scientific Publishing Co. Pte. Ltd, 2006.
[13] Y.-M. Hsieh, M.-H. Bai, J. S. Chang, and K.-J. Chen. Improving pcfg chinese parsing with context-dependent probability reestimation. Proceedings of CLP’12, pp. 216-221. 2012.
[14] Chinese Knowledge Information Processing Group Academia Sinica Institute of Information Science. Technical Report no. 93-05 中文詞類分析(三版)
[15] Chinese Knowledge Information Processing Group Academia Sinica Institute of Information Science. Technical Report no. 13-01 句結構樹中的語意角色
[16] Hoang, Q.: Predicting movie genres based on plot summaries. arXiv preprint arXiv:1801.04813, 2018.
[17] A. M. Ertugrul, P. Karagoz, Movie Genre Classification from Plot Summaries using Bidirectional LSTM, 12th IEEE International Conference on Semantic Computing (ICSC), 2018.
[18] Justin Martineau and Tim Finin. 2009. Delta tfidf: an improved feature space for sentiment analysis. InProceedings of the Third Annual Conference on We-blogs and Social Media, pages 258–261.
[19] Andrew L. Maas, Raymond E. Daly, Peter T. Pham, Dan Huang, Andrew Y. Ng, and Christopher Potts. 2011. Learning word vectors for sentiment analysis. In Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, Vol 1, pages 142–150.
[20] Nguyen DQ, Nguyen DQ, Vu T, Pham SB Sentiment classification on polarity reviews: an empirical study using rating-based features. In: Proceedings of the 5th workshop on computational approaches to subjectivity, sentiment and social media analysis. Baltimore, pp 128–135, 2014.
[21] Maron, M. E. (1961). "Automatic Indexing: An Experimental Inquiry" (PDF). Journal of the ACM. 8 (3): 404–417. doi:10.1145/321075.321084.
[22] Cortes, Corinna; Vapnik, Vladimir N. "Support-vector networks". Machine Learning. 20 (3): 273–297. CiteSeerX 10.1.1.15.9362. doi:10.1007/BF00994018. 1995.
[23] Ho TK . "The Random Subspace Method for Constructing Decision Forests"(PDF). IEEE Transactions on Pattern Analysis and Machine Intelligence. 20 (8): 832–844. doi:10.1109/34.709601, 1998.
[24] Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. BERT: Pre-training of deep bidirectional transformers for language understanding. arXiv preprint arXiv:1810.04805. Google Scholar, 2018.