An Improved Weighted-Removal Sentence Embedding Based Approach for Service Recommendation

Abstract

Currently, there is a large amount of information about user requirements and service in natural language. How to measure the semantic similarity between user requirements and service description is a critical issue in service recommendation and service solution construction. In this paper, we propose a service recommendation method based on the improved Weighted-Removal(WR) sentence embedding to solve the shortcomings of traditional information retrieval methods. After data preprocessing, we use the GloVe method to obtain the word vectors and use the improved WR sentence embedding method to obtain the sentence vectors. The similarity between the vectors can be better measured. The experimental results show that the proposed improved WR method is significantly better than the traditional methods in terms of recommendation accuracy, richness, and ranking.