GeoRSA: Geospatial Recommendation System using Sentimental Analysis

Because of the large amount of information available about locations points of interest on the internet and social media platforms, selecting and visiting places has become difficult for tourist. The opinion, reviews, or feedback of the users about location is available on social media. The feedback of users shared in the form of textual data needs to be analyzed using sentimental analysis. Sentiment analysis can categorize the textual data into different manners like negative, positive, neutral, etc. The sentiment analysis would be applied to both user comments or reviews about the location and location features or characteristics. Then after the sentiment analysis is over, the recommendation or suggestion of the user will take place. This paper discusses the sentiment analysis and different models used or recommendation models in geospatial recommendation systems. Here Bi-LSTM provides nearly 84.7 % accuracy with minimum mean square error (MSE) value as compared to other methods whereas Random Forest methods obtains 85% accuracy with an MSE value of 0.87.