How does one parse XML files?

It's very simple. I know these are standard methods, but you can create your own library to deal with that much better.

Here are some examples:

XmlDocument xmlDoc= new XmlDocument(); // Create an XML document object
xmlDoc.Load("yourXMLFile.xml"); // Load the XML document from the specified file

// Get elements
XmlNodeList girlAddress = xmlDoc.GetElementsByTagName("gAddress");
XmlNodeList girlAge = xmlDoc.GetElementsByTagName("gAge"); 
XmlNodeList girlCellPhoneNumber = xmlDoc.GetElementsByTagName("gPhone");

// Display the results
Console.WriteLine("Address: " + girlAddress[0].InnerText);
Console.WriteLine("Age: " + girlAge[0].InnerText);
Console.WriteLine("Phone Number: " + girlCellPhoneNumber[0].InnerText);

Also, there are some other methods to work with. For example, here. And I think there is no one best method to do this; you always need to choose it by yourself, what is most suitable for you.

I'd use LINQ to XML if you're in .NET 3.5 or higher.

Use a good XSD Schema to create a set of classes with xsd.exe and use an XmlSerializer to create a object tree out of your XML and vice versa. If you have few restrictions on your model, you could even try to create a direct mapping between you model classes and the XML with the Xml*Attributes.

There is an introductory article about XML Serialisation on MSDN.

Performance tip: Constructing an XmlSerializer is expensive. Keep a reference to your XmlSerializer instance if you intend to parse/write multiple XML files.

If you're processing a large amount of data (many megabytes) then you want to be using XmlReader to stream parse the XML.

Anything else (XPathNavigator, XElement, XmlDocument and even XmlSerializer if you keep the full generated object graph) will result in high memory usage and also a very slow load time.

Of course, if you need all the data in memory anyway, then you may not have much choice.