/* @getRecommendations:
* This method implements the Hybrid recommender system, first it finds most similar users by
* checking their tastes (favorite books, musics, movies and athletes) them the next recommender
* algorithm filter users based on Demographic correlation (Location, gender and age), it also features
* their ratings on each product, the output will be a list with the items most appealing to the user */
@RequestMapping(value = "/recommendations", method = RequestMethod.GET)
public ResponseEntity<ArrayList<Prediction>> getRecommendations(@RequestParam("id") String id) {
/* Retrieve all users to cluster */
User baseUser = userRepository.findOne(id);
if (baseUser != null) {
List<User> users = userRepository.findAll();
/* Knowledge Correlation */
ArrayList<Neighbor> knowledgeSimilarity =
ClusterUtils.getKnowledgeNeighborhood(baseUser, users);
List<User> filteredUsers = new ArrayList<User>();
for (int i = 0; i <= (knowledgeSimilarity.size() - 1); i++) {
filteredUsers.add(knowledgeSimilarity.get(i).getUser());
}
ArrayList<Neighbor> neighborhood =
ClusterUtils.getMergedCorrelations(baseUser, filteredUsers);
/* Order arrayList by similar users from highest to lowest correlation */
Collections.sort(neighborhood);
ArrayList<Prediction> predictions = PredictionUtils.getPredictions(baseUser, neighborhood);
Collections.sort(predictions);
return new ResponseEntity<>(predictions, HttpStatus.OK);
} else {
ArrayList<Prediction> predictions = new ArrayList<>();
return new ResponseEntity<>(predictions, HttpStatus.OK);
}
}
@RequestMapping(value = "/trending", method = RequestMethod.GET)
public ResponseEntity<ArrayList<Trending>> getTrending() {
List<User> userList = userRepository.findAll();
List<Occupation> occupationList = occupationRepository.findAll();
ArrayList<Trending> trending = ClusterUtils.getTrendingOccupations(userList, occupationList);
Collections.sort(trending);
return new ResponseEntity<>(trending, HttpStatus.OK);
}