Time series based interactive recommendation for petroleum drilling safety check
Shuang-Bo Sun, Xin-Ling Dong, Zhang Lin, Zhi-Jie Jing, Fan Min · 2016
Recommender systems represent user preferences for the purpose of suggesting items to select or examine. In petroleum drilling safety check, there are many items (e.g. tool misused and warning signs ignored) to be checked during one day. However, existing recommender systems seldom apply time series and interaction methods for the issue. In this paper, we propose a recommender system with two techniques for petroleum drilling safety check. The basic recommender system obtains checking sequence based on statistic average. One of technique is time series prediction. It adopts simple moving average to predict the occurrence of all items in the initial stage. The other technique is user-recommender interaction. It is used to adjust the subsequent checking sequence after interacting with users. Experiments on a real-world petroleum drilling data set show that (1) both techniques help recommender system perform better; and (2) the recommender system with two techniques presents a significant increase than the others.