Update README.md

README.md

@@ -11,4 +11,10 @@ To start analyzing some emails, you can either:
 - Open an existing dataset, via `File > Open Dataset`. You may open datasets from directories, or from compressed ZIP files; in the latter case, the zip file will be unzipped to a directory.
 
 ## Generating a Dataset
+![image](https://user-images.githubusercontent.com/9953867/160172140-2c91753d-d1b1-42ee-8907-ba108360cd68.png)
 
+Datasets are generated by processing one or more directories containing `.mbox` files. After clicking `File > Generate Dataset`, you'll see the above dialog, where you can specify one or more directories to process. If you don't yet have any mbox files, you may click `Download Emails` to open a simple dialog where you can download mbox files from Apache's API by specifying the mailing list name, domain, and a directory to download to. You can browse available mail archives on the [Apache Mail Archives site](https://lists.apache.org/).
+
+Finally, select a directory to generate the dataset in, via the `Generate to` field.
+
+Once you click `Generate`, the mbox files will be parsed to generate a dataset consisting of Lucene index files and an H2 database file, in the specified directory.