DialogueKit is a library for conversational information access (CIA). It contains based classes for fundamental concepts, such as dialogue participants, dialogue management, natural language understanding, natural language generation, etc. In addition to the fundamental concepts DialogueKit contains an evaluation module, for evaluating the performance of and CIA systems. Consult the documentation for details.
DialogueKit is published to PyPI, install it by running:
pip install dialoguekit
Follow the commands below to install DialogueKit from a specific commit or straight from GitHub.
The command will install the latest version from the main branch.
ssh -t git github.com
pip install git+ssh://git@github.com/iai-group/dialoguekit.git
If you want to specify a specific commit as the source of the package append the commit hash to the end of the command separated with a "@".
pip install git+ssh://git@github.com/iai-group/dialoguekit.git@faa5c1fca37aaa275105cc1ca7698783719551c2
Create and connect a platform that manages dialogues. The platform requires information about the agent it serves.
from dialoguekit.platforms.terminal_platform import TerminalPlatform
from sample_agents.parrot_agent import ParrotAgent
platform = TerminalPlatform(ParrotAgent)
Start conversation
platform.start()
We follow the IAI Python Style Guide.
We use UTF-8
encodings that is widely used on Unix systems. Windows users need to use the Python UTF-8 Mode
; see here for more details. In practice, we specify the encoding when opening files, as in this example:
with open("README.md", "r", encoding="utf-8") as fh:
long_description = fh.read()
(Alphabetically ordered by last name)