Big Data Aspect-Based Opinion Mining Using the SLDA and HME-LDA Models

Algorithm 3. The result processing of the Gibbs sampling.

1: //Create a 3d array to store results

2: init results[][][]

3: for topic in TopicList do

4: for wordType in {A, P, N} do

5:  for v in V do

6:   result = new result

7:   result.topic = topic

8:   result.word = v

9:   result.wordType = wordType

10:   Calculate according to Equation. (13), Equation. (14), Equation. (15)

11:   result.prob =

12:   //Add result to the array

13:   results[topic][wordType][v] = result

14:  end for

15: end for

16: end for

17: //Next, sort the results

18: for topic in TopicList do

19: /Store sentences contain results

20: finalResults[][][]

21: for wordType in {A, P, N} do

22:   //Sort byvalue and get the results of the top m

23:   finalResults[topic][wordType] = getTopKByPhi(results[topic][wordType], m)

24:   for to m do

25:   Query the corresponding sentence of result.v from the inverted index described in Section 3.1.1 and add it to the result

26:   end for

27: end for

28: end for

29: return finalResults

Algorithm 3. The result processing of the Gibbs sampling.