The epiWeights function implements the weights-based method. R documentation has a nice introduction to the weights-based method:

> help("epiWeights")

The mechanism of invoking and finally generating the results is very similar to how we did it using emWeights:

library(RecordLinkage)
data("RLdata500")

# weight calculation
rec.pairs <- compare.dedup(RLdata500
                           ,blockfld = list(1, 5:7)
                           ,strcmp =   c(2,3,4)
                           ,strcmpfun = levenshteinSim)

pairs.weights <- epiWeights(rec.pairs)
hist(pairs.weights$Wdata)

One again, the distribution is similar to our histogram from the emWeights method. Let's see an alternate view of the weights distribution:

> summary(pairs.weights)

Deduplication Data Set

500 records 
1221 record pairs 

0 matches
0 non-matches
1221 pairs with unknown status


Weight distribution:

[0.15,0.2] (0.2,0.25] (0.25,0.3] (0.3,0.35] (0.35,0.4] (0.4,0.45] (0.45,0.5] (0.5,0.55] (0.55,0.6] 
       371        445        186         83         66         14          7          1         15 
(0.6,0.65] (0.65,0.7] (0.7,0.75] (0.75,0.8] (0.8,0.85] (0.85,0.9] 
         8         10          0         13          0          2 
> 

Once again we will use getPairs and emClassify as we did in emWeights:

weights.df<-getPairs(pairs.weights)
head(weights.df)

# Classification
pairs.classify <- emClassify(pairs.weights, threshold.upper = 0.5, threshold.lower = 0.3)

# View the matches
final.results <- pairs.classify$pairs
final.results$weight <- pairs.classify$Wdata
final.results$links <- pairs.classify$prediction
head(final.results)

counts <- table(final.results$links)
barplot(counts, main="Link Distribution",
        xlab="Link Types") 

Generate our final list for our customer:

> weights.df.srow <-getPairs( pairs.weights, single.rows = TRUE)
> final.matches <- final.results[final.results$links == 'L',]
> 
> final <- merge(final.matches, weights.df.srow)
> final <- subset(final, select = -c(fname_c1.2, fname_c2.2, lname_c1.2, lname_c2.2, by.2, bm.2, bd.2, weight))
> head(final)
  id1 id2 fname_c1 fname_c2  lname_c1 lname_c2 by bm bd is_match links fname_c1.1 fname_c2.1 lname_c1.1
1 106 175        1       NA 1.0000000       NA  1  0  1       NA     L      ANDRE       <NA>    MUELLER
2 108 203        1       NA 1.0000000       NA  0  1  1       NA     L    GERHARD       <NA>  FRIEDRICH
3 112 116        1       NA 0.8000000       NA  1  1  1       NA     L    GERHARD       <NA>      ERNSR
4 120 165        1       NA 0.8750000       NA  1  1  1       NA     L      FRANK       <NA>   BERGMANN
5 125 193        1       NA 0.8750000       NA  1  1  1       NA     L  CHRISTIAN       <NA>   MUELLEPR
6 127 142        1       NA 0.8333333       NA  1  1  0       NA     L       KARL       <NA>      KLEIN
  lname_c2.1 by.1 bm.1 bd.1    Weight
1       <NA> 1976    2   25 0.6910486
2       <NA> 1987    2   10 0.6133400
3       <NA> 1980   12   16 0.7518301
4       <NA> 1998   11    8 0.7656562
5       <NA> 1974    8    9 0.7656562
6       <NA> 2002    6   20 0.6228760
>