This is a sample Eclipse (and Maven) project for Terrier 5.x, to help an analysis tool (Introductory Lab), a new weighting model (Exercise 1), and later a proximity feature (for Exercise 2).
For the Introductory Lab, you will be expected to edit src/main/java/uk/ac/gla/dcs/applications/HighestTF.java.
In your coursework, following the coursework specification, you will be expected to make new implementations of src/main/java/uk/ac/gla/dcs/models/MyWeightingModel.java (Exercise 1) and src/main/java/uk/ac/gla/dcs/dsms/SampleProxFeatureDSM.java (Exercise 2). You should also edit the corresponding unit tests in src/test/java to ensure your implementations work as you expect.
You can edit/compile/and test your source in Eclipse, IntelliJ or from the command line. You will need to use the command line to deploy your new models within Terrier.
NB: This is supported by the SoCS lab machines.
To import the IRcourseHM project into Eclipse, use File...Import Project...Existing Project from Git (and specify the location of this repository, i.e. https://github.com/cmacdonald/IRcourseHM.git). Alternatively, you can download the IRcourseHM manually, and use the File...Import Project...Existing Maven Project.
You can run the unit tests by selecting Run As...Junit Test from the project's right-click context menu.
You can obtain a jar file at target/ircourse-1.0-SNAPSHOT.jar, by selecting Run As...Maven Build and entering package as the goal. Similarly, you can install that to your local Maven repository by selecting Run As...Maven Build and entering install as the goal.
If you have IntelliJ installed, to import the project, you simply use the File...New...Project from Existing Sources functionality.
You can obtain a jar file at target/ircourse-1.0-SNAPSHOT.jar, by selecting Maven Projects...lifecycle...package on the right hand side of the IntelliJ window. To install that jar file to your local Maven repository, select the install from the lifecycle list. You can run the junit tests by selecting test from the lifecycle list.
If you have Apache Maven installed, you can build and install from the command line. From the command line, in this directory, type:
mvn install
This will create a jar file at target/ircourse-1.0-SNAPSHOT.jar, and will install it into your local Maven repository, so Terrier can find it.
Unit tests are provided for both weighting model and proximity feature DSM. Unit tests are run automatically using mvn install. You can skip unit testing using mvn -DskipTests install.
You can use classes from your project in Terrier, in one of two manners:
- Telling Terrier which Maven package to import.
- manually adding the generate Jar file to Terrier's classpath.
We now explain these two different methods below.
Terrier can search the local Maven repository for packages and add these to the classpath automatically. This assumes that you have used Eclipse/IntelliJ/Maven tools to "install" the package.
Firstly switch (e.g. cd) back to the terrier-platform directory. You will need to specify the group, artifact and version of the package that we want to add to the classpath. There are set in your project's pom.xml file. You add this information to the terrier.mvn.coords property in your terrier.properties file.
terrier.mvn.coords=uk.ac.gla.dcs:ircourse:1.0-SNAPSHOT
Classes defined in this Maven package can now be used in the same manner as Terrier's other standard classes. For instance, to use your new weighting model, specify the name of the class using the -w commandline option:
bin/terrier batchretrieve -w uk.ac.gla.dcs.models.MyWeightingModel
Terrier will tell you that some files have been added to the classpath.
Alternatively, for batchretrieve, this can be done in a single line using the -P commandline option:
bin/terrier batchretrieve -P uk.ac.gla.dcs:ircourse:1.0-SNAPSHOT -w uk.ac.gla.dcs.models.MyWeightingModel
NB: If you change your source code MyWeightingModel, you will need to re-run mvn install for your project.
To use the highest-tf application, ensure that terrier.mvn.coords is appropriately specified in your terrier.properties file as above, then run:
bin/terrier highest-tf
To manually add the generated Jar file to Terrier's classpath, firstly switch (e.g. cd) back to the terrier-platform directory, and then alter the CLASSPATH environment variable to point to the generated jar file, before running bin/terrier. On a Linux machine with a bash shell, this will look like:
CLASSPATH=/path/to/myproject/target/ircourse-1.0-SNAPSHOT.jar bin/terrier batchretrieve
For instance, to use your new weighting model, specify the name of the class using the -w commandline option
CLASSPATH=/path/to/myproject/target/ircourse-1.0-SNAPSHOT.jar bin/terrier batchretrieve -w uk.ac.gla.dcs.models.MyWeightingModel
Or similarly on Windows:
SET CLASSPATH="H:\path\to\myproject\target\ircourse-1.0-SNAPSHOT.jar"
bin\terrier.bat br -Dtrec.model=uk.ac.gla.dcs.models.MyWeightingModel
where br is short for batchretrieve
