complexdb / zincbase Goto Github PK

If you want to start a visualisation on zincbase.web,
you look up the code on basic.py.

Though it is very easy example I couldnt get the whole code because of missing comments.
Commenting would increase the developer experience on zincbase.

Hello, great library! I am making great use of it!

There is a slight issue with the plot() call to networkx.

In zb.py, around line 833, the script returns a networkx error when setting the parameter of font_size.

# line 833 of zb.py
nx.draw_networkx_edges(newg, pos, edgelist=edgelist, width=1, font_size=8)

The line of code above returns an error like the following:

TypeError: draw_networkx_edges() got an unexpected keyword argument 'font_size'

In my local, simply removing the font_size param from this call results in a successful networkx graph of my knowledge base.

# this works
nx.draw_networkx_edges(newg, pos, edgelist=edgelist, width=1)

I am not 100% sure, but based on the networkx function signature, it seems that the font_size param is not supported for draw edges.

If I try to dump a medium complex kb with pickle, I get the exception

`'NoneType' object is not callable

A short google search shows me, that not every class can be dumped. It looks like
the knowledge base class should support this feature.`

I cant find any documentation about zincbase.web.

I would appreciate if the module would be documented.

Hello!

I noticed that the graphing feature, by default, presents a networkx graph of the nodes and edges; however, the spacing can become a bit tight in the layout.

For either a feature enhancement or for anyone interested in spacing out the edges and nodes, consider the following modification to zb.py

   # currently around line 829 and 830 ...
    def plot(self, density=1.0):
        """Plots a network diagram from (triple) nodes and edges in the KB.

        :param float density: Probability (0-1) that a given edge will be plotted, \
        useful to thin out dense graphs for visualization."""
        edgelist = [e for e in self.G.edges(data=True) if random.random() < density]
        newg = nx.DiGraph(edgelist)
        pos = nx.spring_layout(newg)
        # ...

The code above leads to a dense plot as provided by plot1.png

Whereas the code below leads to some more spacing for the figure in plot2.png.

If changed to:

    # code extract to demonstrate enhanced layout
    def plot(self, density=1.0):
        """Plots a network diagram from (triple) nodes and edges in the KB.

        :param float density: Probability (0-1) that a given edge will be plotted, \
        useful to thin out dense graphs for visualization."""
        edgelist = [e for e in self.G.edges(data=True) if random.random() < density]
        newg = nx.DiGraph(edgelist)
        # modification here!
        # this creates some spacing to make tight clusters more readable and could be configured further with 
        # values for K and seed
        pos = nx.spring_layout(newg, k=5 / math.sqrt(newg.order()), seed=0)
        # ...

Hi,

I stumbled upon module not found error in MacOS 10.14.3 (Mojave).

I have tried to install graph-tools in python3 and graph-tool deps via Homebrew.

❯ python3 examples/countries.py
Traceback (most recent call last):
  File "examples/countries.py", line 8, in <module>
    from zincbase import KB
  File "/usr/local/lib/python3.7/site-packages/zincbase/__init__.py", line 1, in <module>
    from .zincbase import KB
  File "/usr/local/lib/python3.7/site-packages/zincbase/zincbase.py", line 20, in <module>
    from graph import Node
ModuleNotFoundError: No module named 'graph'

Any idea how to make it work?