Fragen in Vorstellungsgesprächen für Data Scientist

This question can be answered in two ways - from a use case point of view or by giving a technical explanation. Clustering means finding groups that have a small within-group variance and a high between-group variance. By transferring this concept to text mining, you would want to find clusters of similar documents of any sort. For example this could be clusters of sentiments regarding an event among social media comments ("What do (different groups of) people think about event X?") To give a more technical explanation, the key to this question is extracting a vector of all words from the corpus of all documents. This vector is then used to build a Document Term Matrix, that lists how often each term in the vector is present in each Document. Once you have the Document Term Matrix, you can use your favorite distance measure and clustering algorithm. For example you can use the Euclidean Distance to quantify the difference between documents and applythe k-Means algorithm to the distance matrix. Before you proceed to calculate the distance, you could also further process the Document Term Matrix, for example to compute TF-IDF. I did not talk about data cleansing since this is not the core of this question. https://en.wikipedia.org/wiki/Latent_semantic_analysis https://www.experfy.com/blog/k-means-clustering-in-text-data