# The CSV data source This data source connects CSV files from the web to Prolog predicates. The `csv` specification accepts two arguments: 1. The *URL* to load the data from 2. An option list to control the conversion. Currently this processes: - encoding(+Encoding) Set the encoding for processing the data. Default is guessed from the URL header or `utf8` if there is no clue. - column_names(+ColumnNames) Names for the columns. If not provided, the first row is assumed to hold the column names. ## Example Below we attach the _IRIS dataset_. See also the [iris](example/iris.swinb) notebook, which uses R to perform machine learning on the iris dataset.