To install, run pip install -r requirements.txt in a new virtual environment, and memgraph.sh to start the database. Then, run python main.py to start the chatbot.
There are several tasks to be finished:
- Implement conversations where user input has multiple replies from different bots (think imageboard style threads). We can also do things like fork & join them.
- Represent world states as nodes, and actions as edges. This can be used for planning in a game or similar applications.
- Represent semantic knowledge as a graph, including relations between users, social networks etc.
- Add an index on the vector embedding property and provide an example of searching with it.
- Manage trees of actions, similar to Voyager, for use in planning.
- Relate all the above subgraphs together. For example, check semantic knowledge from some entity in the world state.
After starting Memgraph, go to localhost:3000 to access the visualiser. You can run queries from there. Here's probably the most useful ones::
- Print all nodes:
MATCH (n) RETURN n - Drop graph & start fresh:
MATCH (n) DETACH DELETE n