Add draft description of the project.

README.md

@@ -0,0 +1,33 @@
+# The Graphene
+
+`Graphene` is a header-only library that implements an abstract graph. The class template allows
+creation of graphs of any types of nodes both directed and undirected with
+the custom weight function.
+
+[![Latest release](https://img.shields.io/github/v/release/vahancho/graphene?include_prereleases)](https://github.com/vahancho/graphene/releases)
+[![Build and test (CMake)](https://github.com/vahancho/graphene/actions/workflows/cmake.yml/badge.svg)](https://github.com/vahancho/graphene/actions/workflows/cmake.yml)
+[![Code coverage](https://codecov.io/gh/vahancho/graphene/branch/master/graph/badge.svg)](https://codecov.io/gh/vahancho/graphene)
+
+## Prerequisites
+
+No special requirements except *C++17* compliant compiler. The class is tested with
+*gcc* and *MSVC* compilers. In order to build and run unit tests
+for this project you are required to have Google Test library installed on the system.
+For more details see the CI badges or [GitHub Actions](https://github.com/vahancho/graphene/actions).
+
+## Shortest paths
+
+The library calculates shortest path between two nodes using the Dijkstra algorithm.
+However the term "shortest" is an abstraction, because the distance between
+to nodes is just a particular notion of the weight of the graph edge. `Graphene`
+allows to define custom weight functions, so that the algorithm will calculate
+not only the shortest path, but the path that corresponds to the minimal overall
+"weight".
+
+## API and usage
+
+## Example (California road network)
+
+# See also
+
+[California Road Network](https://users.cs.utah.edu/~lifeifei/SpatialDataset.htm)