User Interest Feature Extraction and Recognition Based on LBS

The description, extraction and identification of user interest characteristics based on LBS can be used to build a location-dependent user interest model [1] to provide location-based personalized advertising referral services. Based on the data provided by Baidu map advertising department, the user data of LBS is used to describe and extract the user's interest. The feature weighting algorithm is designed to improve the interest point recognition model and dig the user's point of interest. The experiment is based on the method designed in this paper from the map to obtain the 100,000 data. The experimental results show that compared with the improved SVM recognition model, this paper improves the SVM feature weighted point recognition model, adds the weight to the original feature, and improves the accuracy of the test data from 70.8% to 72.1%. And further introduces the search result feature on the basis of the feature-weighted SVM, the accuracy of the test data is further increased to 87.3%.