Analysis of the Relevance between Title of Product and Search Term

Abstract

In recent years, the product recommendation algorithm of e-commerce platforms has become more and more important. In this paper, we built a Random Forest Regression model for the problem of predicting the correlation of “search term” and “product title We found this model for the dataset by describing and attributing the products. In the process of numerically calculating features, we applied two types of feature engineering. The first method is to describe and attribute the numbers to the number of words or the length of the sentence. The second method is to use string similarity characteristics to calculate the distance between “search term” and “product title On the results, we got a similar histogram of the correlation scores between the training dataset and the results from the test dataset. The RMSE of the relevance between the training dataset and the predicted value is 0.3179 which indicates that the model is working well.