Long and Short Term User Preference Learning for Next POI Recommendation in deep learning

I need implementation in python or Matlab using a neural network...
Given a comprehensive recommendation system, which has the aptitude to learn users’ long
and short-term preference for the next Point of Interest recommendation.
• Evaluate the model performance for both cold and non-cold start users together to overcome the data sparsity problem arise due to cold-start users.
• Cold-start users are those with little historic data i.e. having little check-in records.
• The probability estimation heavily relies on other users (non-cold-start users).
• A combined technique will be used in this work to understand user check-in behavior at
certain place and time learning contextual features of (POIs) and to mine long and short
term user preference together [6].
• Both these long and short-term behavior are necessary for capturing user preference for the next POI.
in mostly paper data set takes from
1 https://foursquare.com/
2 https://blog.gowalla.com/
3 https://www.yelp.com
i recommended .gowalla dataset
i want to do this task for my Theses my last accuracy is 80% now i want to improve it to more 3 to 7 persentage