Update README.md

README.md

@@ -36,8 +36,10 @@ attracted the documents to the dense area are the topic words.
 >Documents will be placed to other similar documents and close to most distinguishing words. 
 ![Joint Document and Word Embedding](images/doc_word_embedding.svg)
 2. Create lower dimensional embedding of document vectors using [UMAP](https://github.com/lmcinnes/umap)
+>Document vectors in high dimensional space are very sparse, dimension reduction allows the discovery of dense areas. 
 ![UMAP dimension reduced Documents](images/umap_docs.png)
 3. Find dense areas of documents using [HDBSCAN](https://github.com/scikit-learn-contrib/hdbscan)
+>The colored areas are the dense areas of documents. Red points are outliers that do not belong to a specific topic.
 ![HDBSCAN Document Clusters](images/hdbscan_docs.png)
 4. For each dense area calculate centroid of document vectors in original dimension. (centroid = topic vector)
 5. Find n-closest word vectors to the resulting topic vector