01	package com.liangtee.slopeone.recommender.impl;
02	 
03	import java.util.HashMap;
04	import java.util.List;
05	import java.util.Map;
06	import java.util.Set;
07	 
08	import org.slf4j.Logger;
09	import org.slf4j.LoggerFactory;
10	 
11	import com.liangtee.slopeone.recommender.DataSetModel;
12	import com.liangtee.slopeone.recommender.Recommender;
13	import com.liangtee.slopeone.vo.RecommendedItem;
14	 
15	/**
16	 * Weighted Slope One Algorithm
17	 *
18	 * Just for Fun
19	 *
20	 * @author LiangTE
21	 *
22	 */
23	 
24	public class SlopeOneRecommender implements Recommender {
25	 
26	    private static final Logger log = LoggerFactory.getLogger(SlopeOneRecommender.class);
27	     
28	    private DataSetModel dataSet = null;
29	     
30	    public SlopeOneRecommender(DataSetModel dataSet) {
31	        this.dataSet = dataSet;
32	    }
33	 
34	    @Override
35	    public float preferenceOnItem(long userID, long ItemID) throws Exception {
36	         
37	        if(!dataSet.containsUser(userID)) {
38	            throw new Exception("not contains this User : " + userID);
39	        }
40	         
41	        Set<Long> UIDs = dataSet.getUserIDs();
42	        Map<Long, Float> targetUserRatings = dataSet.getUserRatings(userID);
43	        Set<Long> ratedItems = targetUserRatings.keySet();
44	         
45	        Map<Long, Float> diffCache = new HashMap<Long, Float>();
46	        Map<Long, Integer> counter = new HashMap<Long, Integer>();
47	         
48	        log.info("Slope One start to work...");
49	         
50	        for(long UID : UIDs) {                                                  //计算每个用户的差值
51	            if(UID != userID) {
52	                Map<Long, Float> ratings = dataSet.getUserRatings(UID);
53	                for(long iid : ratedItems) {                                    // iid = item id
54	                    if(ratings.containsKey(ItemID) && ratings.containsKey(iid)) {
55	                        float diff = ratings.get(ItemID) - ratings.get(iid);
56	                        if(diffCache.containsKey(iid)) {
57	                            diffCache.put(iid, diffCache.get(iid)+diff);
58	                        } else {
59	                            diffCache.put(iid, diff);
60	                        }
61	                        if(counter.containsKey(iid)) {
62	                            counter.put(iid, counter.get(iid)+1);
63	                        } else {
64	                            counter.put(iid, 1);
65	                        }
66	                    }
67	                }
68	            }
69	             
70	        }
71	         
72	        int K = 0;
73	        float total = 0F;
74	        for(long iid : ratedItems) {
75	            if(diffCache.containsKey(iid)) {
76	                float bias = diffCache.get(iid)/counter.get(iid);
77	                float targetUserRating = targetUserRatings.get(iid);
78	                float p = targetUserRating + bias;
79	                total += p;
80	                K++;
81	            }
82	        }
83	         
84	         
85	        return total/K;
86	    }