Trailhead

Introduction

Trailhead merged with myEdu aims to make it easier to browse course. The code consists of a user interface written in html and javascript and a backend written in scala. The backend is split into two separate parts:

• web: Takes care of requests from the frontend
• paper: Takes care of parsing and analyzing text documents

Setup for testing

For Trailhead to work you must have an instance of mongoDB running in the background and sbt installed on your system. Once you've installed these, clone the repository:

git clone https://github.com/Traill/myedu.git

Next step is to parse a set of documents and add their information to the database. To do so, run

sbt console

Once sbt has compiled the project and opened the console you type the following:

import paper._
val a = Analyzer.initialize("myeducs").parse.link.save

In the last command the following things happen:

• initialize("myeducs") specifies the location of the documents relative the resources folder
• parse takes each document and parses it
• link goes through the documents and calculates the similarity between each document
• save saves the documents to the database

Now that the database is set up, all that is left to do is to write exit to exit the console, and type sbt run from the command line. Trailhead should now be running on localhost

The Analyzer

The Analyzer takes care of tasks to do with parsing texts and measuring their similarities. It consists of the Analyzer case class and a companion object, the Analyzer object, both are in src/paper/Analyzer.scala. Depending on the type of data we are using, we extend analyzer with different traits (MyEdu, ITA2013, ISIT2012). These traits provide the function parseDoc(d : Document) : Document which takes care of getting information out of a file and inserting it in the Document case class (src/paper/Document.scala).

The analyzer provides a few top level functions for handling the data. Starting with the Analyzer companion object we have the following functions:

• initialize(path : String) : Analyzer -- Given a path relative to the resources folder, this will read in the file names of the folder as id's ready to be parsed.
• fromCache(c : String) : Analyzer -- If data has been loaded from folder "my_folder" and saved to Cache, it can be loaded directly by calling Analyzer.fromCache("my_folder").

The Analyzer class then supports the following methods:

• parse : Analyzer -- For every id in the analyzer, the corresponding file is parsed and loaded into the Document case class. Note that you should call Analyzer.initialize before calling parse.
• link : Analyzer -- For every document in the analyzer, the similarity to all other documents is calculated. Note that you should call Analyzer.parse or Analyzer.fromCache("folder_name") before loading since we need the data parsed before it can be linked.
• save : Analyzer -- Save all the documents to the database. This will overwrite previously saved documents from the same folder name without notifying you.
• load : Analyzer -- Load all the documents from the database. Usually used in conjunction with Analyzer.fromCache("folder_name") (e.g.
  Analyzer.fromCache("myedu").load).
• graph : Graph -- Returns a graph structure where each document is a node and the links have weights corresponding to the similarity between two documents. Note that this function doesn't have a return type of Analyzer.
• spectral(k : Int) : Analyzer -- Partitions the graph of documents into k partitions, that are then saved with the Document.
• louvain(treshold : Int = 20) : Analyzer -- As above but using a modularity optimizing graph clustering technique commonly known as the louvain method.
• get(id : String) : Option[Document] -- Returns the document corresponding to the id if such a document exists.

Loading files from another folder

Often while development it will desirable to load files from a folder containing specific documents or fewer files and consequently display them in the web interface. To do so, create a folder in the resources/ directory, say "resources/gnuf" (the '/' is a directory separator. Substitute by the appropriate symbol on non-unix systems). Then to add data to the database, open the sbt console (by running sbt console in the root directory) and execute the following commands:

import paper._
val a = Analyzer.initialize("gnuf").parse.link.save

Once this is done, open the file "src/web/Server.scala" and modify line 15 so "myeducs" is changed to "gnuf". Now you can run the server by calling "sbt run" from the command line and seeing the result in your browser here.

Credits

Trailhead has been developed by Yannik Messerli, Amine Mansour, Jonny Quarta and Jonas Arnfred under the guidance of Rudiger Urbanke.