Installation Steps

Downloading LingPipe

LingPipe is available on the web:

Installing LingPipe

LingPipe needs to be downloaded and unpacked and it's ready to use.

Building and Developing with LingPipe

This section explains how to develop with the LingPipe API, and also how to compile and test the build from source.

Installing Java

LingPipe requires a 1.5 or greater version of Java.

Installing Ant

Ant only needs to be downloaded and installed if you are using it for development and testing.

Installing LingPipe

Windows Compression

For help, see the section below:

Downloading LingPipe

The first step is to download. This can be done from the:

Unpacking LingPipe

Unpack the gzipped tar file into a new directory, which we will henceforth call LingPipeDir.

Installing Java

LingPipe requires a Java 1.4 or greater JVM installation to run and the JDK to compile. If necessary, see the:

Running Demos

GUI Demos

LingPipe runs demos in a graphical user interface (GUI) demos out of the box. For details, see:

Command-Line Demos

Windows Command Line

For help, see the section below:

LingPipe also runs demos through a command-line interrface. In this way, LingPipe can be used in applications without any Java coding. For details, see:

Developing with LingPipe

API Documentation

Complete API documentation is available at:

LingPipe provides all of its functionality in the form of an application program interface (API). Much of this functionality is described in the form of tutorials. For details:

Please note that some links in the tutorials will not work unless the javadoc has been created. See below for instructions on generating the javadoc.

Compiling LingPipe

LingPipe is distributed with a build script for Ant. Although LingPipe may be compiled directly, we highly recommend installing the Apache Ant build tool; see

LingPipe is built from the directory it was unpacked in, LingPipeDir, by invoking a single Ant build target: 

> cd LingPipeDir/trunk
> ant compile

Ant knows, by means of a file naming convention, to invoke LingPipe's top-level build file, build.xml.

Compilation produces the directory LingPipeDir/build/classes, which contains the compiled LingPipe classes (as well as some other resources).

We also recommend that you generate the javadoc: 

> cd LingPipeDir/trunk
> ant javadoc

List of Ant Targets

The additional Ant development targets are listed in the following table.

Ant Targets
Target Description
clean Removes all of the automatically generated components from the distribution. These can be regenerated through the Ant targets described below. This target removes the following automatically generated items:
  • the build directory,
  • the top-level lingpipe-3.9.3.jar LingPipe jar file,
  • the javadoc generated in docs/api,
  • all automatically generated content in the demos by recursively calling the target clean in demos/build.xml.
There are ant targets to recreate each of these.
compile Compiles a version of LingPipe to the build/classes directory. This target does not recursively build the demos.
javadoc Generated the javadoc for LingPipe and places the index in docs/api/index.html from the distribution.
jar Generates a jar (library of compiled code) for LingPipe and puts it in the top level file lingpipe-3.9.3.jar. To actually use the new jar in the distribution, it should be moved up to the top level parallel to the build file. This target will compile if necessary. Note that this will overwrite the existing jar distributed with LingPipe.
test Runs the LingPipe unit tests. Testing is carried out against the top level jar file. This target will compile and create the jar if necessary.
test1 Runs a unit test on a single class. This target will compile and create the jar if necessary. Invoke with system property test.class set to the name of LingPipe package and class to test. For example: 
     ant test1
     -Dtest.class=classify.ClassifierEvaluatorTest
The target will prefix com.aliasi.test.unit and postfix Test, invoking the test case com.aliasi.test.unit.classify.ClassifierEvaluatorTest for the above example.
tar Creates the LingPipe tarball for distribution. It will be created in the file build/lingpipe-3.9.3.tar.gz. This target will create the jar and the web site if necessary.
web Generates the public LingPipe web pages as published at http://alias-i.com/lingpipe. These will be created in the directory docs/api and the home page will be created in docs/api/index.html. This target will create the tarball if necessary, as it is included in the web page.
cd Generates the public LingPipe distribution CD, as distributed at all your favorite conferences. This task will generate the web site if necessary, as it is included with the distribution.

Warning: The Java Runtime Engines are not distributed with LingPipe. In order to create the CD image, the following paths need to exist with the correct files in them for the relevant JRE installers:
  • LingPipe/cd/jre/win/j2re-1_4_2_08-windows-i586-p.exe
  • LingPipe/cd/jre/linux/j2re-1_4_2_08-linux-i586.bin
See the Installing Java section of this document for information on where to find these files on Sun's download pages.

Installing Java

Server versus Client JVM

There are two virtual machines distributed by Sun, the server and client. The server JVM is much faster for larger jobs because it does on-the-fly code unfolding and optimization based on runtime performance statistics.

32-bit JRE distributions only contains the client virtual machine; 32-bit JDK distributions contain both the client and server VM.

The 64-bit JRE and JDK distributions only contain the server VM.

To test if the server version is installed: 

> java -server -version

We recommend installing one of the latest versions of Sun's Java 2 Standard Edition (J2SE). The exact install depends on operating system and intended usage.

Downloading Java

The active versions of Java are:

Java 5 is reaching end-of-life October 30, 2009, and Java 7 has not yet been released as of October 2009, so we recommend version 6.

Checking the Java Install

To test that an appropriate Java is installed, or the instructions below worked, run the following command: 

> java -version

The result should be a version numbered 1.5.0 or higher.

Java End User Install

End users only need the Java 2 Standard Edition (J2SE) Java Runtime Engine (JRE). It contains everything you need to run LingPipe. You will need to choose the appropriate installation for your platform. Make sure it's installed by following the instructions above.

Java Developer Install

Developers will need to install the Java 2 Standard Edition (J2SE) Software Development Kit (SDK). You will need to choose the appropriate installation for your platform.

Java Documentation Install

Sun bundles their Java API documentation separately and you will also need to download and unzip it if you want to read it. You can test it with a web browser.

Installing Ant

Ant is an Apache project used for scripting Java builds. Quoting from Ant's home page, "[Ant] is kind of like Make, but without Make's wrinkles". LingPipe is distributed with a complete set of Ant build files.

The Ant home page is on the Apache site:

The download page is here:

This is set up like most Apache pages to automatically mirror in random rotation. About halfway down the page you will see a bulleted list of hot links for the actual download; e.g. the gzipped tar version is linked as apache-ant-version-bin.tar.gz.

Some of the demos may also require additional jars, such as ones to run Tomcat web deploys, etc. Many of these are distributed in the demos/lib subdirectory of LingPipe.

Uncompressing in Windows: Zip et al.

If you are a Windows user and don't have a utility for untarring, un-gzipping and un-zipping, we would recommend one of the following packages:

The Windows Command Interpreter

Windows can be used from the command line using the DOS shell. There are two standard ways to start it. From the start menu: 

Start >> Programs >> Accessories >> Command Prompt

Or from the Run box, which can be invoked from the start menu: 

Start >> Run

or by holding down the windows key (if you have one) and the letter r. When the Run window pops up, enter cmd into the text box labeled Open: and press the OK button.

The size of the window can be changed by right clicking on its top title bar and selecting Properties. Font size can also be changed. Lucida Console is good for coding because it's clear and narrow. We would also suggest the quick edit and insert modes from the options menu.

For more information see: