利用attention机制,把word embeddings生成sentence embeddings,再利用sentence embedding生成story embedding。
同理,把演员表集合中导演embedding生成Cast embeddings,
把以上两者并接起来(concatenate)
因为一个时间演员集合,一个是文本描述。所以是Multimodal.
本文绘图很紧致;提出了一个在softmax 之前改变因变量的函数ISR;本文在学出权重(attention)后,还visualization了结果。比如用深颜色表达权重大的words或setences
文献题目 | 去谷歌学术搜索 | ||||||||||
Interpretable Recommendation via Attraction Modeling: Learning Multilevel Attractiveness over Multimodal Movie Contents | |||||||||||
文献作者 | Liang Hu | ||||||||||
文献发表年限 | 2018 | ||||||||||
文献关键字 | |||||||||||
attention; 绘图 | |||||||||||
摘要描述 | |||||||||||
New contents like blogs and online videos are produced in every second in the new media age. We argue that attraction is one of the decisive factors for user selection of new contents. However, collaborative filtering cannot work without user feedback; and the existing content-based recommender systems are ineligible to capture and interpret the attractive points on new contents. Accordingly, we propose attraction modeling to learn and interpret user attractiveness. Specially, we build a multi-level attraction model (MLAM) over the content features—the story (textual data) and cast members (categorical data) of movies. In particular, we design multilevel personal filters to calculate users’ attractiveness on words, sentences and cast members at different levels. The experimental results show the superiority of MLAM over the state-of-the-art methods. In addition, a case study is provided to demonstrate the interpretability of MLAM by visualizing user attractiveness on a movie. |