Questions and AnswersCategory: Text Mining QuestionCould I use k-means instead of RxNLP Sentence Clustering?
Vik asked 9 months ago

I have been using k-means for sentence clustering. I just wanted to know what the benefit is of using RxNLP Sentence Clustering?

1 Answers
Best Answer
KG Staff answered 9 months ago

Yes, you can certainly use k-means for sentence clustering. K-means does not provide approximate labels for your clusters. Also it generates hard clusters (a single sentence could only belong to one cluster) unless you use the Fuzzy version of k-means. The RxNLP Sentence Clustering however uses a soft-clustering approach and provides topic cues for each of the clusters. So you will roughly understand why a cluster was formed. For example, in the cluster below you know that the cluster was formed because of “terrorist attack” and “attack”


{
                "clusterScore": 1.4905037742070442,
                "clusterSize": 3,
                "clusterTopics": "[terrorist attack:1.49, attack:1.02]",
                "clusteredSentences": [
                    "0000: The terrorist attack was really horrifying",
                    "0001: People were really afraid after the terrorist attack",
                    "0002: The terrorists escaped after the attacks "
                ]
            },