From textual descriptions of People and sequence (before, after, not first) get the potential orderings.
Using python-constraint for defining and solving the problem, and ply for lexing and parsing textual descriptions.
See the tutorial for a detailed view of the syntax.
See releases at present only a Windows build has been made.
This will create an temporary playground in Google Cloud for you to see the application in action. It does require a Google Cloud account, as of August 2019 there is no charge for using Google Cloud Shell (link).
Assuming:
- Python 3.6+ installed
virtualenvinstalled
I've created a go script for Windows and Mac. The go script will:
- Create a virtual python environment.
- Install the requirements.
- Run the tests.
- Run the application.
$ ./go_macosx.sh
May need to $ chmod 755 go_macosx.sh
go_windows.bat
The following enhancements have been identified and ranked using the tool. The above image is an example of the output that can be generated by the tool.
Why? Make it easier for humans to specify things.
What? At present entities with spaces are only supported between square brackets. Future state would be to assume all words between tokens that match a token pattern make up the entity name.
Now: [Orange cat] before [Grey cat]
Later: Orange cat before Grey cat
Why? Penwidth on exported diagrams is too large, the graph is an unreadable mess of lines.
What When generating the graphviz export, normalise the range of penwidths between 1 and 20 (range is just an estimate). While preserving the ratio of differences.
Why? I can't spell to save my life.
What? Allowing for [Project Management] and [Project Managment] to resolve to the same thing. Perhaps with a "did you mean workflow".
Why? At the moment all querying is done in the same "space". This means you either create a mega-context or start from scratch every time.
What? Allow the user to state and switch between projects. Perhaps projects take the form of a directory which can store outputs and be used to regenerate state between query sessions.
Why? Graph generation is presently limited to the creation of a graphviz textual description of a graph, which the user has to then manually turn in a diagram (https://dreampuf.github.io/GraphvizOnline/ - has been my goto).
What? Generate the diagram, this would require extra dependencies on the part of the environment.
Why? When solving a diagram with few constraints the answer isn't very helpful.
What? When few specified constraints, prompt the user with some options perhaps using a pairwise ranking?
Why? Foundations of moving this out of a single user environment (e.g. consolse) to something more modern (e.g. Web, bots).
What Some form of user identification and session isolation?