KGCN II - Relation prediction using DeepMind's Graph Nets (#85)

## What is the goal of this PR?

Create a more versatile approach to KGCNs, one that can be customised to perform:
- Relation prediction
- Attribute value prediction
- Subgraph prediction
- Graph optimisation
- Generation of Concept embeddings

## What are the changes implemented in this PR?

Overwrites the entirety of the KGCN project with a more advanced version. This new approach uses a learned message-passing algorithm, implemented in TensorFlow. The algorithm can predict node, edge and global properties for any input graph.

Implemented is a method to predict new Relations in extracted subgraphs, picking from a provided set of candidate relations.

Solves #51, #78, #42

.circleci/config.yml

@@ -9,7 +9,24 @@ commands:
             chmod +x bazel-0.25.2-installer-linux-x86_64.sh
             sudo ./bazel-0.25.2-installer-linux-x86_64.sh
 
+  run-grakn-server:
+    steps:
+      - bazel_install
+      - run: bazel build @graknlabs_grakn_core//:assemble-linux-targz
+      - run: mkdir dist && tar -xvzf bazel-genfiles/external/graknlabs_grakn_core/grakn-core-all-linux.tar.gz -C ./dist/
+      - run: nohup ./dist/grakn-core-all-linux/grakn server start
+
 jobs:
+  build:
+    machine: true
+    working_directory: ~/kglib
+    steps:
+      - checkout
+      - bazel_install
+      - run: pyenv install 3.6.3
+      - run: pyenv global 3.6.3
+      - run: bazel build //...
+
   test:
     machine: true
     working_directory: ~/kglib
@@ -41,19 +58,29 @@ jobs:
         - checkout
         - run: pyenv install 3.6.3
         - run: pyenv global 3.6.3
-        - run: wget https://storage.googleapis.com/kglib/grakn-core-all-mac-animaltrade1.5.3.zip
-        - run: unzip grakn-core-all-mac-animaltrade1.5.3.zip
-        - run: nohup grakn-core-all-mac-animaltrade1.5.3/grakn server start
+        - run-grakn-server
         - run:
             name: Run test-deployment-pip for kglib
             command: |
               echo -n "$(cat VERSION)-$CIRCLE_SHA1" > VERSION
-              sed -i -e "s/KGLIB_VERSION_MARKER/$(cat VERSION)/g" test/deployment/requirements.txt
-              cat test/deployment/requirements.txt
-              pip install -r test/deployment/requirements.txt
-              python -m unittest examples.kgcn.animal_trade.test.end_to_end_test
+              sed -i -e "s/KGLIB_VERSION_MARKER/$(cat VERSION)/g" tests/deployment/requirements.txt
+              cat tests/deployment/requirements.txt
+              pip install -r tests/deployment/requirements.txt
+              ./dist/grakn-core-all-linux/grakn console -k diagnosis -f kglib/utils/grakn/synthetic/examples/diagnosis/schema.gql
+              python -m unittest kglib.kgcn.examples.diagnosis.diagnosis_end_to_end_test
 
-  deploy-git:
+  release-approval:
+    machine: true
+    working_directory: ~/kglib
+    steps:
+      - checkout
+      - bazel_install
+      - run: |
+          export RELEASE_APPROVAL_USERNAME=$REPO_GITHUB_USERNAME
+          export RELEASE_APPROVAL_TOKEN=$REPO_GITHUB_TOKEN
+          bazel run @graknlabs_build_tools//ci:release-approval
+
+  deploy-github:
     machine: true
     working_directory: ~/kglib
     steps:
@@ -79,28 +106,66 @@ jobs:
           export DEPLOY_PIP_PASSWORD=$REPO_PYPI_PASSWORD
           bazel run //:deploy-pip -- release
 
+  release-cleanup:
+    machine: true
+    steps:
+      - checkout
+      - run: git push --delete origin kglib-release-branch
+
 workflows:
   version: 2
   kglib:
     jobs:
-      - test
+      - test:
+          filters:
+            branches:
+              ignore: kglib-release-branch
+      - build:
+          filters:
+            branches:
+              ignore: kglib-release-branch
       - deploy-pip-snapshot:
+          filters:
+            branches:
+              only: master
           requires:
             - test
+            - build
       - test-deployment-pip:
+          filters:
+            branches:
+              only: master
           requires:
             - deploy-pip-snapshot
-      - approve-deploy-git:
-          type: approval
+      - release-approval:
+          filters:
+            branches:
+              only: master
           requires:
             - test-deployment-pip
-      - deploy-git:
-          requires:
-            - approve-deploy-git
-      - approve-deploy-pip-release:
+
+  kglib-release:
+    jobs:
+      - deploy-github:
+          filters:
+            branches:
+              only: kglib-release-branch
+      - deploy-approval:
           type: approval
+          filters:
+            branches:
+              only: kglib-release-branch
           requires:
-            - deploy-git
+            - deploy-github
       - deploy-pip-release:
+          filters:
+            branches:
+              only: kglib-release-branch
+          requires:
+            - deploy-approval
+      - release-cleanup:
+          filters:
+            branches:
+              only: kglib-release-branch
           requires:
-            - approve-deploy-pip-release
+            - deploy-pip-release

BUILD

@@ -28,12 +28,44 @@ assemble_pip(
     author = "Grakn Labs",
     author_email = "community@grakn.ai",
     license = "Apache-2.0",
-    install_requires=['absl-py==0.5.0', 'astor==0.7.1', 'decorator==4.3.0', 'gast==0.2.0', 'grakn-client==1.5.3', 'h5py==2.8.0',
-                      'Keras-Applications==1.0.6', 'Keras-Preprocessing==1.0.5',
-                      'Markdown==3.0.1', 'networkx==2.2', 'numpy==1.15.2', 'protobuf==3.6.1', 'scikit-learn==0.20.1',
-                      'scipy==1.1.0', 'six==1.11.0', 'tensorboard==1.11.0', 'tensorflow==1.11.0',
-                      'tensorflow-hub==0.1.1', 'termcolor==1.1.0', 'Werkzeug==0.14.1',
-                      'grpcio==1.16.0', 'protobuf==3.6.1', 'six==1.11.0', 'enum34==1.1.6', 'twine==1.12.1', 'requests==2.21.0'],
+    install_requires=[
+        'enum-compat==0.0.2',
+        'grakn-client==1.5.4',
+        'absl-py==0.8.0',
+        'astor==0.8.0',
+        'cloudpickle==1.2.2',
+        'contextlib2==0.5.5',
+        'cycler==0.10.0',
+        'decorator==4.4.0',
+        'dm-sonnet==1.35',
+        'future==0.17.1',
+        'gast==0.3.1',
+        'google-pasta==0.1.7',
+        'graph-nets==1.0.4',
+        'grpcio==1.23.0',
+        'h5py==2.10.0',
+        'Keras-Applications==1.0.8',
+        'Keras-Preprocessing==1.1.0',
+        'kiwisolver==1.1.0',
+        'Markdown==3.1.1',
+        'matplotlib==3.1.1',
+        'networkx==2.3',
+        'numpy==1.17.2',
+        'pandas==0.25.1',
+        'protobuf==3.9.1',
+        'pyparsing==2.4.2',
+        'python-dateutil==2.8.0',
+        'pytz==2019.2',
+        'semantic-version==2.8.2',
+        'six==1.12.0',
+        'tensorboard==1.14.0',
+        'tensorflow==1.14.0',
+        'tensorflow-estimator==1.14.0',
+        'tensorflow-probability==0.7.0',
+        'termcolor==1.1.0',
+        'Werkzeug==0.15.6',
+        'wrapt==1.11.2',
+    ],
     keywords = ["machine learning", "logical reasoning", "knowledege graph", "grakn", "database", "graph",
                 "knowledgebase", "knowledge-engineering"],
 

README.md

@@ -12,6 +12,35 @@ To respond to these scenarios, KGLIB is the centre of all research projects cond
 
 At present this repo contains one project: [*Knowledge Graph Convolutional Networks* (KGCNs)](https://github.com/graknlabs/kglib/tree/master/kglib/kgcn).
 
+## Quickstart
+**Requirements**
+
+- Python 3.7
+
+- Install via pip: `pip install grakn-kglib`
+
+- The [latest release of Grakn Core](https://github.com/graknlabs/grakn/releases/latest) or [Grakn KGMS](https://dev.grakn.ai/docs/cloud-deployment/kgms) running
+
+**Building from source**
+
+To test that all targets can be built: 
+
+```bash
+bazel build //...
+```
+
+To run all tests: 
+
+```bash
+bazel test //... --test_output=streamed --spawn_strategy=standalone --python_version PY3 --python_path $(which python3)
+```
+
+To build the pip distribution (find the output in `bazel-bin`):
+
+```bash
+bazel build //:assemble-pip
+```
+
 ## Knowledge Graph Tasks
 
 Below is a set of tasks to be conducted over Knowledge Graphs (KGs) that we have identified from real Grakn use cases. The objective of KGLIB is to implement a portfolio of solutions for these tasks for Grakn Knowledge Graphs.

VERSION

@@ -1 +1 @@
-0.1.1
+0.2