本文的核心观点:在原来LSTM的基础上，将time interval的信息显性（explicit）的加入到网络结构当中去。这么做的道理：LSTM当中，虽然有序的概念，但没有把时间间隔考虑进去。这个思想用在对句子的刻画当中没有问题（连续两个单词之间不存在时间间隔，只有先后次序），而对于推荐系统而言，用户对物品的访问不仅有先后顺序，还有时间间隔。所以time-LSTM就是有意义的。
具体做法：用时间间隔乘以一个参数矩阵，输入到sigmoid function当中变成权重，这个权重就可以同LSTM当中原有的权重控制一通作用forget掉的信息和新加入的信息。注意文中Eq15中的对于W小于0的约束很有意思。考虑到了离当前时间间隔越近，当前信息/input权重越大的这样一个思想。
有意思的是，本文在LSTM的基础上，更好的融入了时间差信息，并且把这个信息直接加入到了LSTM到网络结果中，即Time-LSTM. 另一篇以相同的方式处理time interval的文章:Dynamic Explainable Recommendation based on Neural Attentive Modelsd
| 文献题目 | 去谷歌学术搜索 | ||||||||||
| What to Do Next: Modeling User Behaviors by Time-LSTM | |||||||||||
| 文献作者 | Yu Zhu | ||||||||||
| 文献发表年限 | 2017 | ||||||||||
| 文献关键字 | |||||||||||
| explicit time interval; time signal; LSTM; Phased LSTM; 融入时间信息； time interval2vec | |||||||||||
| 摘要描述 | |||||||||||
| Recently, Recurrent Neural Network (RNN) solutions for recommender systems (RS) are becoming increasingly popular. The insight is that, there exist some intrinsic patterns in the sequence of users’ actions, and RNN has been proved to perform excellently when modeling sequential data. In traditional tasks such as language modeling, RNN solutions usually only consider the sequential order of objects without the notion of interval. However, in RS, time intervals between users’ actions are of significant importance in capturing the relations of users’ actions and the traditional RNN architectures are not good at modeling them. In this paper, we propose a new LSTM variant, i.e. Time-LSTM, to model users’ sequential actions. Time-LSTM equips LSTM with time gates to model time intervals. These time gates are specifically designed, so that compared to the traditional RNN solutions, Time-LSTM better captures both of users’ short- term and long-term interests, so as to improve the recommendation performance. Experimental results on two real-world datasets show the superiority of the recommendation method using Time- LSTM over the traditional methods. | |||||||||||
