README.md

CONQUEST

CONQUEST (Chatbot ONtology QUESTion) is a framework that automates much of the construction process of chatbots for the task of template-based Interactive Question Answering on closed-domain knowledge bases.

Installation

Automatic Installation

Just run the script "install.sh"
$ ./install.sh

Manual Installation

• Execute the command: $ pip install -r requirements.txt
• Execute the command: $ python -c "import nltk; nltk.download('punkt')"
• Download the file "sorl.tar.xz" from https://sourceforge.net/projects/conquest-sqai/files/solr.tar.xz/download
• Extract the "sorl.tar.xz" file in "persistence/nlp" keeping the name "sorl"
• Download the file "model.tar.xz" from https://sourceforge.net/projects/conquest-sqai/files/model_pt-br.tar.xz/download
• Extract the "model.tar.xz file in "persistence/nlp" keeping the name "model"
• Execute the command: $ python -m spacy download en_core_web_md
• Remove temporary files

Configuration

It is possible to configure the framework through a Web interface using the CONQUEST Workbench:

• $ python WorkBench.py
• Access the link in the Web browser: http://localhost:5050/
• Configure the chatbot parameters and enter the supported templates.

Trainer

The target ontology file should be placed in the directory "input" with the name "ontology.ttl" in the turle serialization.

To run the training script:

$ python Trainer.py

usage: python Trainer.py [-h] [-m zero] [-p 0]
optional arguments:
-h, --help				Show this help message and exit 
-m zero, --mode zero	Training mode: 
						zero - Training the chatbot from starting point (default).
						update - Update models
						resume - Resume the training from a saved point.
-p 0, --point 0			Point to resume training:
							0 - Starting from start (equal to zero mode).
							1 - Starting from loading QAIs.
							2 - Starting from making NER training dataset.
							3 - Starting from training NER. 
							4 - Starting from making classifier training dataset. 
							5 - Starting from training the classifier.
	

Server

To run the server script:

The chatbot service will be available in the link http://0.0.0.0:5000/

$ python Server.py

To run the RDF Browser (Optional):

The RDF Browser will be available in the link: http://0.0.0.0:5555/ and through the functions /explorar $term or /browser $term in the chatbot.

• Enter in the directory RDF_Browser: $ cd RDF_Browser
• Run the command: $ python server.py