FirstAppPipeline

[Michael Matthews]

Adding Analysis

To add analysis to the given collection, you modify the xml configuration file (conf/testbed/tutorial-application.xml).

Text Analysis

The text analysis tools to be run are specified in under the annotators element of the pipeline element. Each tool specified in this list is run sequentially in the order occurring in the list. To add additional analysis tools, you simply add the tools to the list. The tools are documented in the doc directory. This link between the tools and the documentation will be improved in future. If any dependencies are specified in the documentation, they must also be run (this does not yet happen automatically, but an error will be given when the testbed application is run).
For example, to add facts analysis and the Subjective expressions analysis, you would the annotators specified below to the configuration file

    <pipeline>
        <annotators>
....
            <annotator exec="./unitn_tagger"/>
            <annotator exec="./unitn_subjexpr"/>
            <annotator exec="./imageannots"/>
        </annotators>
    </pipeline>

To perform the analysis,you need to re-run the testbed

apps/testbed -run pipeline,visualize conf/testbed/tutorial-application.xml

Notice that this time, we are only running the pipeline and visualize components of the testbed. These are specified by listing the components you want run (separated by commas without spaces) after the -run parameter. You can verify the analysis has been done by noticing the new files in the devapps/example/data/lkxml directory.

Image Analysis

The image analysis tools to be run are specified in under the annotators element of the image-pipeline element. As with text analysis, each tool specified in this list is run sequentially in the order occurring in the list. To add additional analysis tools, you simply add the tools to the list. For example, to add the harr facet detector you would the annotators specified below to the configuration file

    <image-pipeline>
        <annotators>
            <annotator exec="./soton_haarfacedetector" />
        </annotators>
    </image-pipeline>

To perform the analysis,you need to re-run the testbed

apps/testbed -run image-pipeline conf/testbed/tutorial-application.xml

You can verify the analysis has been done by noticing the new files in the devapps/example/data/lkxml/img directory. Currently, there is no visualization for the image output. This may come in future versions.

---

Next: Indexing and Search