Install Linux packages

Table of Contents generated with DocToc

• Installing LIMA on Linux
  • Svmtool++
  • Qhttpserver
  • TensorFlow
  • Install LIMA
    • Install language models

Installing LIMA on Linux

LIMA uses the Travis CI continuous integration service to build new Linux packages for each push to its Github git repository. These packages are then uploaded to GitHub as a new release. Currently (October 2021), packages are built for Debian 11 and Ubuntu 20.04. But these are updated from time to time when new stable version of distributions are released. So, check what is available at the time of your reading.

To install LIMA packages, you have to install some dependencies. Under Ubuntu 20.04, this is done with:

$ sudo apt-get install qt5-default libqt5xmlpatterns5 libqt5quick5 libboost-all-dev libenchant1c2a libtre5 \
python3 libqt5qml5 qml-module-qtquick-extras \
qml-module-qtquick-controls qml-module-qtquick-layouts qml-module-qtquick2 \
qml-module-qtquick-controls2 qml-module-qtquick-dialogs \
qml-module-qtquick-privatewidgets qml-module-qtquick-scene3d qml-module-qtquick-templates2 \
qml-module-qtquick-virtualkeyboard qml-module-qtquick-window2 qml-module-qtquick-xmllistmodel

Svmtool++

Download the latest version of svmtool++.

Install it:

$ sudo dpkg -i svmtool++-1.1.7-ubuntu16.deb

Qhttpserver

Download the latest version of qhttpserver.

Install it:

$ sudo dpkg -i qhttpserver-0.0.1-ubuntu16.deb

TensorFlow

For its neural network based modules, LIMA depends on a specifically built version of TensorFlow. This was necessary due to the fact that by default TensorFlow uses a non-standard build system and do not generate dynamically-linked libraries, which is a problem as soon as we have several plugins using it.

To make easy the installation of our own version of TF, we have set up a dedicated PPA (Ubuntu apt repository). From Ubuntu 18.04 up, you just have to run:

$ sudo apt-get install -y software-properties-common
$ sudo add-apt-repository -y ppa:limapublisher/ppa
$ sudo apt-get update
$ sudo apt-get install -y libtensorflow-for-lima-dev

Install LIMA

Download LIMA package.

And install them. For the Tensor:

$ sudo apt install lima-XXX.deb

LIMA is then ready to be used. Refer to the LIMA user manual for instructions.

Install language models

To use the deep learning based LIMA modules, you must download models. Please refer to the dedicated page to do so (list available languages, install them…). When it is done, you can use them with the deepud pipeline:

$ analyzeText -l ud-eng -p deepud your-text-file.txt