7yl4r / neurosemantics

An attempt at a hybrid between adaptive artificial neural networks and semantic network AI.

neuroSemantics

An attempt at a hybrid between adaptive artificial neural networks and semantic network AI.

The network takes simplified language as input and creates/modifies a network based on given statements. Answers to queries are generated by propagating the semantic meaning of the question through the network and retrieving the 'most activated' nodes. Exact methodology here is still in the works.

The network class allows for the creation of any arbitrarily connected network which can be visualized from a csv linkmap. Current visualization includes a force directed diagram generated using D3.

Attributions

D3 visualization coding based on the work of d3noob

neurosemantics's Issues

improved viz loading

having to manually start-up the HTTP server to make javascript visualizations work is a bit of a pain.

Unfortunately, Popen() is not properly killing the process in del() either.

So it seems that the server must be either manually started or manually stopped... there must be a better way of doing this.

IOError: [Errno 2] No such file or directory: './states/s0.jpg'

Running the main.py file gives the following error:

WARN: range <= 0
conn stren m: 100.0 v: 0
adjustment m: 2900.0 v: 0
Traceback (most recent call last):
File "main.py", line 40, in
network.input_file('./input_scripts/base.txt')
File "/Users/school/Github/neuroSemantics/Network.py", line 175, in input_file
self.plot_state_file()
File "/Users/school/Github/neuroSemantics/Network.py", line 162, in plot_state_file
self.state.plot_img('./states/s' + str(self.input_count) + '.jpg')
File "/Users/school/Github/neuroSemantics/NetworkState.py", line 69, in plot_img
img.save(save_file_loc)
File "/Library/Python/2.7/site-packages/PIL/Image.py", line 1684, in save
fp = builtins.open(filename, "wb")
IOError: [Errno 2] No such file or directory: './states/s0.jpg'

I have looked around for a bit, what exactly am I doing wrong?? I feel like it could be a problem with my PIL installation (installed via pip install pillow )

normalize strengths & state values

Using integers is getting messy, need to switch to using floats between 0-1

implement csv representation of network in network.py

The methods there are just filler for now, they should actually read and write to the file.