Feature Data Selection for Improving the Performance of Entity Similarity Searches in the Internet of Things

Abstract

Sensors are used to sense the state information of physical entities in the Internet of Things (IoT). Thus, a large amount of dynamic real-time data is generated. The entity similarity search based on the quantitative dynamic sensor data is thus worth studying. To the best of our knowledge, there is no research on the entity similarity search based on feature data selection for the quantitative dynamic sensor data in the IoT. This paper proposes a selection mechanism for the entity main features (SMEF). The SMEF is a feature data selection method based on the quantitative dynamic sensor data. It uses the feature matrix to delete the irrelevant entity features, applies an improved relief algorithm (iRelief) to calculate the feature data relevance and proposes a three-component storage relation table of the entities, models, and features (TEMF) for the dynamic feature weights calculation. The experimental results show that the similarity search algorithm based on feature data selection can improve the average search accuracy by more than 10%, as well as increase the search speed and reduce the data transmission and storage costs.

GAWA–A Feature Selection Method for Hybrid Sentiment Classification

“…Sharma et al [35] proposed a hybrid ensemble learning method for the optimal features sets by utilizing various feature selection algorithm. Liu S. and Fan et al [36] proposed a selection method for the entity main features, which relies on the quantitative dynamic sensor data. It utilized the feature matrix to remove the inappropriate entity features and employed iRelief algorithm to compute the relevant features.…”